< draft-ietf-intarea-gue-06.txt   draft-ietf-intarea-gue-06cepC.txt >
skipping to change at page 1, line 48 skipping to change at page 1, line 48
Copyright (c) 2018 IETF Trust and the persons identified as the Copyright (c) 2018 IETF Trust and the persons identified as the
document authors. All rights reserved. document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of (http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect carefully, as they describe your rights and restrictions with respect
to this document. to this document.
<!-- CEP: duplicate boilerplate!! -->
This document is subject to BCP 78 and the IETF Trust's Legal This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of (http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License. described in the Simplified BSD License.
Abstract Abstract
This specification describes Generic UDP Encapsulation (GUE), which This specification describes Generic UDP Encapsulation (GUE), which
is a scheme for using UDP to encapsulate packets of different IP uses UDP to encapsulate packets of various Internet
protocols for transport across layer 3 networks. By encapsulating protocols for transport across layer 3 networks. By encapsulating
packets in UDP, specialized capabilities in networking hardware for packets in UDP, specialized capabilities in networking hardware for
efficient handling of UDP packets can be leveraged. GUE specifies efficient handling of UDP packets can be used. GUE provides
basic encapsulation methods upon which higher level constructs, such basic encapsulation methods suitable for higher level constructs, such
as tunnels and overlay networks for network virtualization, can be as tunnels and overlay networks for network virtualization.
constructed. GUE is extensible by allowing optional data fields as GUE provides extensibility by allowing optional data fields within
part of the encapsulation, and is generic in that it can encapsulate the encapsulation, and is generic in that it can encapsulate
packets of various IP protocols. packets of various Internet protocols.
<!-- CEP: This means GUE is "flexible", not "generic". Oh well. -->
<!-- CEP: "generic" would mean that GUE offers a vanilla solution
relevant to other encapsulation protocols. -->
Table of Contents Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 5 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1. Terminology and acronyms . . . . . . . . . . . . . . . . . 5 1.1. Terminology and acronyms . . . . . . . . . . . . . . . . . 5
1.2. Requirements Language . . . . . . . . . . . . . . . . . . 6 1.2. Requirements Language . . . . . . . . . . . . . . . . . . 6
2. Base packet format . . . . . . . . . . . . . . . . . . . . . . 7 2. Base packet format . . . . . . . . . . . . . . . . . . . . . . 7
2.1. GUE variant . . . . . . . . . . . . . . . . . . . . . . . . 7 2.1. GUE variant . . . . . . . . . . . . . . . . . . . . . . . . 7
3. Variant 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 3. Variant 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1. Header format . . . . . . . . . . . . . . . . . . . . . . . 8 3.1. Header format . . . . . . . . . . . . . . . . . . . . . . . 8
skipping to change at page 4, line 34 skipping to change at page 4, line 34
10.1. Normative References . . . . . . . . . . . . . . . . . . . 30 10.1. Normative References . . . . . . . . . . . . . . . . . . . 30
10.2. Informative References . . . . . . . . . . . . . . . . . . 30 10.2. Informative References . . . . . . . . . . . . . . . . . . 30
Appendix A: NIC processing for GUE . . . . . . . . . . . . . . . . 33 Appendix A: NIC processing for GUE . . . . . . . . . . . . . . . . 33
A.1. Receive multi-queue . . . . . . . . . . . . . . . . . . . . 33 A.1. Receive multi-queue . . . . . . . . . . . . . . . . . . . . 33
A.2. Checksum offload . . . . . . . . . . . . . . . . . . . . . 34 A.2. Checksum offload . . . . . . . . . . . . . . . . . . . . . 34
A.2.1. Transmit checksum offload . . . . . . . . . . . . . . . 34 A.2.1. Transmit checksum offload . . . . . . . . . . . . . . . 34
A.2.2. Receive checksum offload . . . . . . . . . . . . . . . 35 A.2.2. Receive checksum offload . . . . . . . . . . . . . . . 35
A.3. Transmit Segmentation Offload . . . . . . . . . . . . . . . 35 A.3. Transmit Segmentation Offload . . . . . . . . . . . . . . . 35
A.4. Large Receive Offload . . . . . . . . . . . . . . . . . . . 36 A.4. Large Receive Offload . . . . . . . . . . . . . . . . . . . 36
Appendix B: Implementation considerations . . . . . . . . . . . . 36 Appendix B: Implementation considerations . . . . . . . . . . . . 36
B.1. Priveleged ports . . . . . . . . . . . . . . . . . . . . . 37 B.1. Privileged ports . . . . . . . . . . . . . . . . . . . . . 37
B.2. Setting flow entropy as a route selector . . . . . . . . . 37 B.2. Setting flow entropy as a route selector . . . . . . . . . 37
B.3. Hardware protocol implementation considerations . . . . . . 37 B.3. Hardware protocol implementation considerations . . . . . . 37
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 38 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 38
1. Introduction 1. Introduction
This specification describes Generic UDP Encapsulation (GUE) which is This specification describes Generic UDP Encapsulation (GUE) which is
a general method for encapsulating packets of arbitrary IP protocols a general method for encapsulating packets of arbitrary IP protocols
within User Datagram Protocol (UDP) [RFC0768] packets. Encapsulating within User Datagram Protocol (UDP) [RFC0768] packets. Encapsulating
packets in UDP facilitates efficient transport across networks. packets in UDP facilitates efficient transport across networks.
Networking devices widely provide protocol specific processing and Networking devices often provide protocol specific processing and
optimizations for UDP (as well as TCP) packets. Packets for atypical optimizations for UDP packets. Packets for
IP protocols (those not usually parsed by networking hardware) can be IP protocols not typically parsed by networking hardware can be
encapsulated in UDP packets to maximize deliverability and to encapsulated in UDP packets to maximize deliverability and to
leverage flow specific mechanisms for routing and packet steering. engage flow specific mechanisms for routing and packet steering.
GUE provides an extensible header format for including optional data GUE provides an extensible header format for including optional data
in the encapsulation header. This data potentially covers items such in the encapsulation header. This data can cover items such
as the virtual networking identifier, security data for validating or as the virtual networking identifier, security data for validating or
authenticating the GUE header, congestion control data, etc. GUE also authenticating the GUE header, congestion control data, etc.
<!-- CEP: citations needed. Plus I am surprised yet another
validation mechanism is needed! -->
GUE also
allows private optional data in the encapsulation header. This allows private optional data in the encapsulation header. This
feature can be used by a site or implementation to define local feature can be used by a site or implementation to define local
custom optional data, and allows experimentation of options that may custom optional data, and allows experimentation of options that may
eventually become standard. eventually become standard.
This document does not define any specific GUE extensions. [GUEEXTEN] This document does not define any specific GUE extensions. [GUEEXTEN]
specifies a set of initial extensions. specifies a set of initial extensions.
The motivation for the GUE protocol is described in section 6. The motivation for the GUE protocol is described in section 6.
1.1. Terminology and acronyms 1.1. Terminology and acronyms
<!-- CEP: Need terminology for "connection semantics". -->
<!-- CEP: Need terminology for "flow entropy". -->
<!-- CEP: Need terminology for "Canonical length". -->
GUE Generic UDP Encapsulation GUE Generic UDP Encapsulation
GUE Header A variable length protocol header that is composed GUE Header A variable length protocol header that is composed
of a primary four byte header and zero or more four of a primary four byte header and zero or more four
byte words for optional header data byte words for optional header data
GUE packet A UDP/IP packet that contains a GUE header and GUE GUE packet A UDP/IP packet that contains a GUE header and GUE
payload within the UDP payload payload within the UDP payload
GUE variant A version of the GUE protocol or an alternate form GUE variant A version of the GUE protocol or an alternate form
skipping to change at page 7, line 15 skipping to change at page 7, line 15
2. Base packet format 2. Base packet format
A GUE packet is comprised of a UDP packet whose payload is a GUE A GUE packet is comprised of a UDP packet whose payload is a GUE
header followed by a payload which is either an encapsulated packet header followed by a payload which is either an encapsulated packet
of some IP protocol or a control message such as an OAM (Operations, of some IP protocol or a control message such as an OAM (Operations,
Administration, and Management) message. A GUE packet has the general Administration, and Management) message. A GUE packet has the general
format: format:
+-------------------------------+ +-------------------------------+
| | | |
| UDP/IP header | | UDP/IP headers |
| | | |
|-------------------------------| |-------------------------------|
| | | |
| GUE Header | | GUE Header |
| | | |
|-------------------------------| |-------------------------------|
| | | |
| Encapsulated packet | | Encapsulated packet |
| or control message | | or control message |
| | | |
+-------------------------------+ +-------------------------------+
<!-- CEP: This representation seems upside down compared to normal.
Maybe even horizontal would be better... -->
The GUE header is variable length as determined by the presence of The GUE header has variable length, as determined by the presence of
optional extension fields. optional extension fields.
2.1. GUE variant 2.1. GUE variant
The first two bits of the GUE header contain the GUE protocol variant The first two bits of the GUE header contain the GUE protocol variant
number. The variant number can indicate the version of the GUE number. The variant number can indicate the version of the GUE
protocol as well as alternate forms of a version. protocol as well as alternate forms of a version.
Variants 0 and 1 are described in this specification; variants 2 and Variants 0 and 1 are described in this specification; variants 2 and
3 are reserved. 3 are reserved.
3. Variant 0 3. Variant 0
Variant 0 indicates version 0 of GUE. This variant defines a generic Variant 0 indicates version 0 of GUE, and defines a generic
extensible format to encapsulate packets by Internet protocol number. extensible format to encapsulate packets by Internet protocol number.
3.1. Header format 3.1. Header format
The header format for variant 0 of GUE in UDP is: The header format for variant 0 of GUE in UDP is:
<!-- CEP: Is it possible to have a GUE *not* in UDP?
If not, then the wording above is curious... -->
0 1 2 3 0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+\ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+\
| Source port | Destination port | | | Source port | Destination port | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ UDP +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ UDP
| Length | Checksum | | | Length | Checksum | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+/ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+/
| 0 |C| Hlen | Proto/ctype | Flags | | 0 |C| Hlen | Proto/ctype | Flags |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
skipping to change at page 8, line 45 skipping to change at page 8, line 47
to an encapsulation, this is set to the local source port for to an encapsulation, this is set to the local source port for
the connection. When connection semantics are not applied, the the connection. When connection semantics are not applied, the
source port is either set to a flow entropy value as described source port is either set to a flow entropy value as described
in section 5.11, or it should be set to the GUE assigned port in section 5.11, or it should be set to the GUE assigned port
number, 6080. number, 6080.
o Destination port: If connection semantics (section 5.6.1) are o Destination port: If connection semantics (section 5.6.1) are
applied to an encapsulation, this is set to the destination port applied to an encapsulation, this is set to the destination port
for the tuple. If connection semantics are not applied this is for the tuple. If connection semantics are not applied this is
set to the GUE assigned port number, 6080. set to the GUE assigned port number, 6080.
<!-- CEP: Why not always use the former? -->
o Length: Canonical length of the UDP packet (length of UDP header o Length: Canonical length of the UDP packet (length of UDP header
and payload). and payload).
o Checksum: Standard UDP checksum (handling is described in o Checksum: Standard UDP checksum (handling is described in
section 5.7). section 5.7).
The GUE header consists of: The GUE header consists of:
o Variant: 0 indicates GUE protocol version 0 with a header. o Variant: 0 indicates GUE protocol version 0 with a header.
o C: C-bit: When set indicates a control message, not set o C: C-bit: When set indicates a control message, not set
indicates a data message. indicates a data message.
o Hlen: Length in 32-bit words of the GUE header, including o Hlen: Length in 4-byte words of the GUE header, including
optional extension fields but not the first four bytes of the optional extension fields but not the first four bytes of the
header. Computed as (header_len - 4) / 4, where header_len is header. Computed as (header_len - 4) / 4, where header_len is
the total header length in bytes. All GUE headers are a multiple the total header length in bytes. All GUE headers are a multiple
of four bytes in length. Maximum header length is 128 bytes. of four bytes in length. Maximum header length is 128 bytes.
o Proto/ctype: When the C-bit is set, this field contains a o Proto/ctype: When the C-bit is set, this field contains a
control message type for the payload (section 3.2.2). When the control message type for the payload (section 3.2.2). When the
C-bit is not set, the field holds the Internet protocol number C-bit is not set, the field holds the Internet protocol number
for the encapsulated packet in the payload (section 3.2.1). The for the encapsulated packet in the payload (section 3.2.1). The
control message or encapsulated packet begins at the offset control message or encapsulated packet begins at the offset
provided by Hlen. provided by Hlen.
o Flags: Header flags that may be allocated for various purposes o Flags: Header flags that may be allocated for various purposes
and may indicate presence of extension fields. Undefined header and may indicate presence of extension fields. Undefined header
flag bits MUST be set to zero on transmission. flag bits MUST be set to zero on transmission.
o Extension Fields: Optional fields whose presence is indicated by o Extension Fields: Optional fields whose presence is indicated by
corresponding flags. corresponding flags.
o Private data: Optional private data block (see section 3.4). If o Private data: Optional private data block (see section 3.4). If
the private block is present, it immediately follows that last the private block is present, it immediately follows the last
extension field present in the header. The private block is extension field present in the header. The private block is
considered to be part of the GUE header. The length of this data considered to be part of the GUE header. The length of this data
is determined by subtracting the starting offset from the header is determined by subtracting the starting offset from the header
length. length.
<!-- CEP: not clear why it's better to be part of the GUE header, than
another kind of extension field. -->
3.2. Proto/ctype field 3.2. Proto/ctype field
The proto/ctype fields either contains an Internet protocol number The proto/ctype fields either contains an Internet protocol number
(when the C-bit is not set) or GUE control message type (when the C- (when the C-bit is not set) or GUE control message type (when the C-
bit is set). bit is set).
3.2.1 Proto field 3.2.1 Proto field
When the C-bit is not set, the proto/ctype field MUST contain an IANA When the C-bit is not set, the proto/ctype field MUST contain an IANA
Internet Protocol Number. The protocol number is interpreted relative Internet Protocol Number. The protocol number is interpreted relative
to the IP protocol that encapsulates the UDP packet (i.e. protocol of to the IP protocol that encapsulates the UDP packet (i.e. protocol of
the outer IP header). The protocol number serves as an indication of the outer IP header).
<!-- CEP: Presumably this means IPv4 or IPv6, but it's the same for
either of those. -->
The protocol number indicates
the type of the next protocol header which is contained in the GUE the type of the next protocol header which is contained in the GUE
payload at the offset indicated in Hlen. Intermediate devices MAY payload at the offset indicated in Hlen. Intermediate devices MAY
parse the GUE payload per the number in the proto/ctype field, and parse the GUE payload per the number in the proto/ctype field, and
header flags cannot affect the interpretation of the proto/ctype header flags MUST NOT affect the interpretation of the proto/ctype
field. field.
<!-- CEP: There are three possibilities below. They should be itemized. -->
When the outer IP protocol is IPv4, the proto field MUST be set to a When the outer IP protocol is IPv4, the proto field MUST be set to a
valid IP protocol number usable with IPv4; it MUST NOT be set to a valid IP protocol number usable with IPv4. An
number for IPv6 extension headers or ICMPv6 options (number 58). An
exception is that the destination options extension header using the exception is that the destination options extension header using the
PadN option MAY be used with IPv4 as described in section 3.6. The PadN option MAY be used with IPv4 as described in section 3.6. The
"no next header" protocol number (59) also MAY be used with IPv4 as "no next header" protocol number (59) also MAY be used with IPv4 as
described below. described below.
<!-- CEP: should cite the website for IP protocol numbers:
https://www.iana.org/assignments/protocol-numbers/protocol-numbers.xhtml
-->
When the outer IP protocol is IPv6, the proto field can be set to any When the outer IP protocol is IPv6, the proto field can be set to any
defined protocol number except that it MUST NOT be set to Hop-by-hop defined protocol number except that it MUST NOT be set to Hop-by-hop
options (number 0). If a received GUE packet in IPv6 contains a options (number 0).
<!-- CEP: Please explain the rationale for this restriction. -->
If a received GUE packet in IPv6 contains a
protocol number that is an extension header (e.g. Destination protocol number that is an extension header (e.g. Destination
Options) then the extension header is processed after the GUE header Options) then the extension header is processed after the GUE header
is processed as though the GUE header is an extension header. is processed as though the GUE header is an extension header.
IP protocol number 59 ("No next header") can be set to indicate that IP protocol number 59 ("No next header") can be set to indicate that
the GUE payload does not begin with the header of an IP protocol. the GUE payload does not begin with the header of an IP protocol.
This would be the case, for instance, if the GUE payload were a This would be the case, for instance, if the GUE payload were a
fragment when performing GUE level fragmentation. The interpretation fragment when performing GUE level fragmentation. The interpretation
of the payload is performed through other means (such as flags and of the payload is performed through other means (such as flags and
extension fields), and intermediate devices MUST NOT parse packets extension fields), and intermediate devices MUST NOT parse packets
based on the IP protocol number in this case. based on the IP protocol number in this case.
<!-- CEP: This cannot be enforced. -->
3.2.2 Ctype field 3.2.2 Ctype field
When the C-bit is set, the proto/ctype field MUST be set to a valid When the C-bit is set, the proto/ctype field MUST be set to a valid
control message type. A value of zero indicates that the GUE payload control message type. A value of zero indicates that the GUE payload
requires further interpretation to deduce the control type. This requires further interpretation to deduce the control type. This
might be the case when the payload is a fragment of a control might be the case when the payload is a fragment of a control
message, where only the reassembled packet can be interpreted as a message, where only the reassembled packet can be interpreted as a
control message. control message.
Control messages will be defined in an IANA registry. Control message Control messages are defined in an IANA registry. Control message
types 1 through 127 may be defined in standards. Types 128 through types 1 through 127 may be defined in standards. Types 128 through
255 are reserved to be user defined for experimentation or private 255 are reserved to be user defined for experimentation or private
control messages. control messages.
<!-- CEP: For types 1 --> 127, need to specify how to allocate.
Why not mandate standards action? -->
This document does not specify any standard control message types This document does not specify any standard control message types
other than type 0. Type 0 does not define a format of the control other than type 0. Type 0 does not define a format of the control
message. Instead, it indicates that the GUE payload is a control message. Instead, it indicates that the GUE payload is a control
message, or part of a control message (as might be the case in GUE message, or part of a control message (as might be the case in GUE
fragmentation), that cannot be correctly parsed or interpreted fragmentation), that cannot be correctly parsed or interpreted
without additional context. without additional context.
<!-- CEP: The latter needs an example. The former seems to follow
from network mishandling of IPv6 fragmentation. -->
3.3. Flags and extension fields 3.3. Flags and extension fields
Flags and associated extension fields are the primary mechanism of Flags and associated extension fields are the primary mechanism of
extensibility in GUE. As mentioned in section 3.1, GUE header flags extensibility in GUE. As mentioned in section 3.1, GUE header flags
indicate the presence of optional extension fields in the GUE header. indicate the presence of optional extension fields in the GUE header.
[GUEXTENS] defines an initial set of GUE extensions. [GUEEXTENS] defines an initial set of GUE extensions.
3.3.1. Requirements 3.3.1. Requirements
There are sixteen flag bits in the GUE header. Flags may indicate There are sixteen flag bits in the GUE header. Flags may indicate
presence of an extension fields. The size of an extension field presence of an extension fields. The size of an extension field
indicated by a flag MUST be fixed. indicated by a flag MUST be a fixed constant.
<!-- CEP: "paired" means 2. Need another term. Perhaps "combined". -->
Flags can be paired together to allow different lengths for an Flags can be paired together to allow different lengths for an
extension field. For example, if two flag bits are paired, a field extension field. For example, if two flag bits are paired, a field
can possibly be three different lengths-- that is bit value of 00 can possibly be three different lengths-- that is bit value of 00
indicates no field present; 01, 10, and 11 indicate three possible indicates no field present; 01, 10, and 11 indicate three possible
lengths for the field. Regardless of how flag bits are paired, the lengths for the field. Regardless of how flag bits are paired, the
lengths and offsets of optional fields corresponding to a set of lengths and offsets of optional fields corresponding to a set of
flags MUST be well defined. flags MUST be well defined.
Extension fields are placed in order of the flags. New flags are to Extension fields are placed in order of the flags. New flags are to
be allocated from high to low order bit contiguously without holes. be allocated from high to low order bit contiguously without holes.
Flags allow random access, for instance to inspect the field Flags allow random access, for instance to inspect the field
corresponding to the Nth flag bit, an implementation only considers corresponding to the Nth flag bit, an implementation only considers
the previous N-1 flags to determine the offset. Flags after the Nth the previous N-1 flags to determine the offset. Flags after the Nth
flag are not pertinent in calculating the offset of the field for the flag are not pertinent in calculating the offset of the field for the
Nth flag. Random access of flags and fields permits processing of Nth flag. Random access of flags and fields permits processing of
optional extensions in an order that is independent of their position optional extensions in an order that does not depend on their position
in the packet. in the packet.
Flags (or paired flags) are idempotent such that new flags MUST NOT Flags (or paired flags) are idempotent such that new flags MUST NOT
<!-- CEP: This is not what "idempotent" means. -->
cause reinterpretation of old flags. Also, new flags MUST NOT alter cause reinterpretation of old flags. Also, new flags MUST NOT alter
interpretation of other elements in the GUE header nor how the interpretation of other elements in the GUE header nor how the
message is parsed (for instance, in a data message the proto/ctype message is parsed (for instance, in a data message the proto/ctype
field always holds an IP protocol number as an invariant). field always holds an IP protocol number as an invariant).
The set of available flags can be extended in the future by defining The set of available flags can be extended in the future by defining
a "flag extensions bit" that refers to a field containing a new set a "flag extensions bit" that refers to a field containing a new set
of flags. of flags.
<!-- CEP: The extension bit should be specified in this document. -->
3.3.2. Example GUE header with extension fields 3.3.2. Example GUE header with extension fields
An example GUE header for a data message encapsulating an IPv4 packet An example GUE header for a data message encapsulating an IPv4 packet
and containing the Group Identifier and Security extension fields and containing the Group Identifier and Security extension fields
(both defined in [GUEXTENS]) is shown below: (both defined in [GUEEXTENS]) is shown below:
0 1 2 3 0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| 0 |0| 3 | 94 |1|0 0 1| 0 | | 0 |0| 3 | 94 |1|0 0 1| 0 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Group Identifier | | Group Identifier |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| | | |
+ Security + + Security +
| | | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
<!-- CEP: The 'C' bit should be a 1, right? Otherwise the 94
has to be an IPv6 protocol number. -->
In the above example, the first flag bit is set which indicates that In the above example, the first flag bit is set which indicates that
the Group Identifier extension is present which is a 32 bit field. the Group Identifier extension is present which is a 32 bit field.
The second through fourth bits of the flags are paired flags that The second through fourth bits of the flags are combined flags that
indicate the presence of a Security field with seven possible sizes. indicate the presence of a Security field with seven possible sizes.
In this example 001 indicates a sixty-four bit security field. In this example 001 indicates a sixty-four bit security field.
<!-- CEP: This seems to make the GUEEXTENS document normative. -->
3.4. Private data 3.4. Private data
An implementation MAY use private data for its own use. The private An implementation MAY use private data for its own use.
<!-- CEP: This sentence does not seem helpful. -->
The private
data immediately follows the last field in the GUE header and is not data immediately follows the last field in the GUE header and is not
a fixed length. This data is considered part of the GUE header and a fixed length. This data is considered part of the GUE header and
MUST be accounted for in header length (Hlen). The length of the MUST be accounted for in header length (Hlen). The length of the
private data MUST be a multiple of four and is determined by private data MUST be a multiple of four and is determined by
subtracting the offset of private data in the GUE header from the subtracting the offset of private data in the GUE header from the
header length. Specifically: header length. Specifically:
Private_length = (Hlen * 4) - Length(flags) Private_length = (Hlen * 4) - Length(extensions)
where "Length(flags)" returns the sum of lengths of all the extension where "Length(extensions)" returns the sum of lengths of all the extension
fields present in the GUE header. When there is no private data fields following the GUE header. When there is no private data
present, the length of the private data is zero. present, the length of the private data is zero.
The semantics and interpretation of private data are implementation The semantics and interpretation of private data are implementation
specific. The private data may be structured as necessary, for specific. An encapsulator and decapsulator MUST agree on the meaning of
instance it might contain its own set of flags and extension fields. private data before using it. The mechanism to achieve this agreement is
outside the scope of this document.
An encapsulator and decapsulator MUST agree on the meaning of private
data before using it. The mechanism to achieve this agreement is
outside the scope of this document but could include implementation-
defined behavior, coordinated configuration, in-band communication
using GUE control messages, or out-of-band messages.
If a decapsulator receives a GUE packet with private data, it MUST If a decapsulator receives a GUE packet with private data, it MUST
validate the private data appropriately. If a decapsulator does not validate the private data. If a decapsulator does not
expect private data from an encapsulator, the packet MUST be dropped. expect private data from an encapsulator, the packet MUST be dropped.
If a decapsulator cannot validate the contents of private data per If a decapsulator cannot validate the contents of private data per
the provided semantics, the packet MUST also be dropped. An the provided semantics, the packet MUST also be dropped.
<!-- CEP: since the structure of the private data is out of scope,
the following RFC 2119 language is used incorrectly.
An
implementation MAY place security data in GUE private data which if implementation MAY place security data in GUE private data which if
present MUST be verified for packet acceptance. present MUST be verified for packet acceptance. -->
3.5. Message types 3.5. Message types
3.5.1. Control messages 3.5.1. Control messages
Control messages carry formatted data that are implicitly addressed Control messages carry formatted data that are implicitly addressed
to the decapsulator to monitor or control the state or behavior of a to the decapsulator to monitor or control the state or behavior of a
tunnel (OAM). For instance, an echo request and corresponding echo tunnel. For instance, an echo request and corresponding echo
reply message can be defined to test for liveness. reply message can be defined to test for liveness.
Control messages are indicated in the GUE header when the C-bit is Control messages are present in the GUE header when the C-bit is
set. The payload is interpreted as a control message with type set. The payload is interpreted as a control message with type
specified in the proto/ctype field. The format and contents of the specified in the proto/ctype field. The format and contents of the
control message are indicated by the type and can be variable length. control message are indicated by the type and can be variable length.
Other than interpreting the proto/ctype field as a control message Other than interpreting the proto/ctype field as a control message
type, the meaning and semantics of the rest of the elements in the type, the meaning and semantics of the rest of the elements in the
GUE header are the same as that of data messages. Forwarding and GUE header are the same as that of data messages. Forwarding and
routing of control messages should be the same as that of a data routing of control messages should be the same as that of a data
message with the same outer IP and UDP header and GUE flags; this message with the same outer IP and UDP header and GUE flags; this
ensures that control messages can be created that follow the same ensures that control messages can be created that follow the same
skipping to change at page 13, line 49 skipping to change at page 13, line 52
packet of an Internet protocol indicated in the proto/ctype field. packet of an Internet protocol indicated in the proto/ctype field.
The packet immediately follows the GUE header. The packet immediately follows the GUE header.
3.6. Hiding the transport layer protocol number 3.6. Hiding the transport layer protocol number
The GUE header indicates the Internet protocol of the encapsulated The GUE header indicates the Internet protocol of the encapsulated
packet. A protocol number is either contained in the Proto/ctype packet. A protocol number is either contained in the Proto/ctype
field of the primary GUE header or in the Payload Type field of a GUE field of the primary GUE header or in the Payload Type field of a GUE
Transform extension field (used to encrypt the payload with DTLS, Transform extension field (used to encrypt the payload with DTLS,
[GUEEXTEN]). If the transport protocol number needs to be hidden from [GUEEXTEN]). If the transport protocol number needs to be hidden from
the network, then a trivial destination options can be used. the network, then a trivial destination options can be used, as
specified below.
<!-- CEP: This destination option needs to be specified in this document.-->
The PadN destination option [RFC2460] can be used to encode the The PadN destination option [RFC2460] can be used to encode the
<!-- CEP: PadN is not a destination option. -->
transport protocol as a next header of an extension header (and transport protocol as a next header of an extension header (and
maintain alignment of encapsulated transport headers). The maintain alignment of encapsulated transport header). The
Proto/ctype field or Payload Type field of the GUE Transform field is Proto/ctype field or Payload Type field of the GUE Transform field is
set to 60 to indicate that the first encapsulated header is a set to 60 to indicate that the first encapsulated header is a
destination options extension header. destination options extension header.
The format of the extension header is below: The format of the extension header is below:
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Next Header | 2 | 1 | 0 | | Next Header | 2 | 1 | 0 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
For IPv4, it is permitted in GUE to used this precise destination For IPv4, it is permitted in GUE to used this precise destination
option to contain the obfuscated protocol number. In this case next option to hide the protocol number. In this case next
header MUST refer to a valid IP protocol for IPv4. No other extension header MUST refer to a valid IP protocol for IPv4. No other extension
headers or destination options are permitted with IPv4. headers or destination options are permitted with IPv4.
4. Variant 1 4. Variant 1
Variant 1 of GUE allows direct encapsulation of IPv4 and IPv6 in UDP. Variant 1 of GUE allows direct encapsulation of IPv4 and IPv6 in UDP.
In this variant there is no GUE header; a UDP packet carries an IP In this variant there is no GUE header; a UDP packet carries an IP
packet. The first two bits of the UDP payload for GUE are the GUE packet. The first two bits of the UDP payload for GUE are the GUE
variant and coincide with the first two bits of the version number in variant and coincide with the first two bits of the version number in
the IP header. The first two version bits of IPv4 and IPv6 are 01, so the IP header. The first two version bits of IPv4 and IPv6 are 01, so
skipping to change at page 17, line 27 skipping to change at page 17, line 27
| Host 1 | | Host 2 | | Host 1 | | Host 2 |
| | | | | | | |
+---------------+ +---------------+ +---------------+ +---------------+
| ^ | ^
V | V |
+---------------+ +---------------+ +---------------+ +---------------+ +---------------+ +---------------+
| | | | | | | | | | | |
| Encapsulator |-->| Layer 3 |-->| Decapsulator | | Encapsulator |-->| Layer 3 |-->| Decapsulator |
| | | Network | | | | | | Network | | |
+---------------+ +---------------+ +---------------+ +---------------+ +---------------+ +---------------+
<!-- CEP: This is true for any L3 encapsulation -->
<!-- CEP: This figure illustrates something that is most likely
obvious to almost all readers. -->
The encapsulator and decapsulator may be co-resident with the The encapsulator and decapsulator may be co-resident with the
corresponding hosts, or may be on separate nodes in the network. corresponding hosts, or may be on separate nodes in the network.
5.1. Network tunnel encapsulation 5.1. Network tunnel encapsulation
Network tunneling can be achieved by encapsulating layer 2 or layer 3 Network tunneling can be achieved by encapsulating layer 2 or layer 3
packets. In this case the encapsulator and decapsulator nodes are the packets. In this case the encapsulator and decapsulator nodes are the
tunnel endpoints. These could be routers that provide network tunnels tunnel endpoints. These could be routers that provide network tunnels
on behalf of communicating hosts. on behalf of communicating hosts.
<!-- CEP: Do you mean that GUE can encapsulate L2 frames? If so, this
contradicts earlier text in several places. --->
5.2. Transport layer encapsulation 5.2. Transport layer encapsulation
When encapsulating layer 4 packets, the encapsulator and decapsulator When encapsulating layer 4 packets, the encapsulator and decapsulator
should be co-resident with the hosts. In this case, the encapsulation should be co-resident with the hosts.
<!-- CEP: This seems to try to distinguish between "layer 4" and
IP protocol number. If so, what is the distinction?? -->
In this case, the encapsulation
headers are inserted between the IP header and the transport packet. headers are inserted between the IP header and the transport packet.
The addresses in the IP header refer to both the endpoints of the The addresses in the IP header refer to both the endpoints of the
encapsulation and the endpoints for terminating the transport encapsulation and the endpoints for terminating the transport
protocol. Note that the transport layer ports in the encapsulated protocol. Note that the transport layer ports in the encapsulated
packet are independent of the UDP ports in the outer packet. packet are independent of the UDP ports in the outer packet.
Details about performing transport layer encapsulation are discussed Details about performing transport layer encapsulation are discussed
in [TOU]. in [TOU].
<!-- CEP: I think those details belong here. If not, this is a normative
dependency on a non-WG document that is not yet mature. -->
5.3. Encapsulator operation 5.3. Encapsulator operation
Encapsulators create GUE data messages, set the fields of the UDP Encapsulators create GUE data messages, set the fields of the UDP
header, set flags and optional extension fields in the GUE header, header, set flags and optional extension fields in the GUE header,
and forward packets to a decapsulator. and forward packets to a decapsulator.
<!-- CEP: According to the figure, encapsulators don't have to
create data. Moreover, the whole point of "generic"
encapsulation seems to be independence from content. -->
An encapsulator can be an end host originating the packets of a flow, An encapsulator can be an end host originating the packets of a flow,
or can be a network device performing encapsulation on behalf of or can be a network device performing encapsulation on behalf of
hosts (routers implementing tunnels for instance). In either case, hosts (routers implementing tunnels for instance). In either case,
the intended target (decapsulator) is indicated by the outer the intended target (decapsulator) is indicated by the outer
destination IP address and destination port in the UDP header. destination IP address and destination port in the UDP header.
If an encapsulator is tunneling packets -- that is encapsulating If an encapsulator is tunneling packets -- that is encapsulating
packets of layer 2 or layer 3 protocols (e.g. EtherIP, IPIP, ESP packets of layer 3 protocols (e.g. EtherIP, IPIP, ESP
tunnel mode) -- it SHOULD follow standard conventions for tunneling tunnel mode) -- it MUST follow standard conventions for tunneling
of one protocol over another. For instance, if an IP packet is being of one protocol over another. For instance, if an IP packet is being
encapsualated in GUE then diffserv interaction [RFC2983] and ECN encapsulated in GUE then diffserv interaction [RFC2983] and ECN
propagation for tunnels [RFC6040] SHOULD be followed. propagation for tunnels [RFC6040] MUST be followed.
<!-- CEP: this violates "generic" unless all IP protocols are
respected in that way. -->
<!-- CEP: GUE states several times that implementations MUST follow
standards. Does that mean that sometimes implementations
MAY break standards if not explicitly required to follow
them? I hope not! I think it would be better to avoid
recommendations about following existing standards. -->
5.4. Decapsulator operation 5.4. Decapsulator operation
A decapsulator performs decapsulation of GUE packets. A decapsulator A decapsulator performs decapsulation of GUE packets. A decapsulator
is addressed by the outer destination IP address of a GUE packet. is addressed by the outer destination IP address of a GUE packet.
The decapsulator validates packets, including fields of the GUE The decapsulator validates packets, including fields of the GUE
header. header.
If a decapsulator receives a GUE packet with an unsupported variant, If a decapsulator receives a GUE packet with an unsupported variant,
unknown flag, bad header length (too small for included extension unknown flag, bad header length (too small for included extension
fields), unknown control message type, bad protocol number, an fields), unknown control message type, bad protocol number, an
unsupported payload type, or an otherwise malformed header, it MUST unsupported payload type, or an otherwise malformed header, it MUST
drop the packet. Such events MAY be logged subject to configuration drop the packet. Such events MAY be logged subject to configuration
and rate limiting of logging messages. Note that set flags in a GUE and rate limiting of logging messages. Note that set flags in a GUE
header that are unknown to a decapsulator MUST NOT be ignored. If a header that are unknown to a decapsulator MUST NOT be ignored. If a
GUE packet is received by a decapsulator with unknown flags, the GUE packet is received by a decapsulator with unknown flags, the
packet MUST be dropped. packet MUST be dropped.
<!-- CEP: An ICMP message seems appropriate here. -->
5.4.1. Processing a received data message 5.4.1. Processing a received data message
If a valid data message is received, the UDP header and GUE header If a valid data message is received, the UDP header and GUE header
<!-- CEP: Need to define "valid". -->
are removed from the packet. The outer IP header remains intact and are removed from the packet. The outer IP header remains intact and
the next protocol in the IP header is set to the protocol from the the next protocol in the IP header is set to the protocol from the
proto field in the GUE header. The resulting packet is then proto field in the GUE header. The resulting packet is then
resubmitted into the protocol stack to process that packet as though resubmitted into the protocol stack to process that packet as though
it was received with the protocol in the GUE header. it was received with the protocol in the GUE header.
As an example, consider that a data message is received where GUE As an example, consider that a data message is received where GUE
encapsulates an IPv4 packet using GUE variant 0. In this case proto encapsulates an IPv4 packet using GUE variant 0. In this case proto
field in the GUE header is set to 4 for IPv4 encapsulation: field in the GUE header is set to 4 for IPv4 encapsulation:
+-------------------------------------+ +-------------------------------------+
| IP header (next proto = 17,UDP) | | IP header (next proto = 17,UDP) |
|-------------------------------------| |-------------------------------------|
| UDP | | UDP |
|-------------------------------------| |-------------------------------------|
| GUE (proto = 4,IPv4 encapsulation) | | GUE (proto = 4,IPv4 encapsulation) |
|-------------------------------------| |-------------------------------------|
| IPv4 header and packet | | IPv4 header and packet |
+-------------------------------------+ +-------------------------------------+
<!-- CEP: All figures should have figure numbers, even if not
cross-referenced in this document. -->
The receiver removes the UDP and GUE headers and sets the next The receiver removes the UDP and GUE headers and sets the next
protocol field in the IP packet to 4, which is derived from the GUE protocol field in the IP packet to 4, which is derived from the GUE
proto field. The resultant packet would have the format: proto field. The resultant packet would have the format:
+-------------------------------------+ +-------------------------------------+
| IP header (next proto = 4,IPv4) | | IP header (next proto = 4,IPv4) |
|-------------------------------------| |-------------------------------------|
| IP header and packet | | IP header and packet |
+-------------------------------------+ +-------------------------------------+
This packet is then resubmitted into the protocol stack to be This packet is then resubmitted into the protocol stack to be
processed as an IPv4 encapsulated packet. processed as an IPv4 encapsulated packet.
5.4.2. Processing a received control message 5.4.2. Processing a received control message
If a valid control message is received, the packet MUST be processed If a valid control message is received, the packet MUST be processed
as a control message. The specific processing to be performed depends as a control message.
<!-- CEP: Need to define "valid". What exactly is meant by "processed
as a control message"? In each case, the processing is
determined by the GUE header fields. -->
The specific processing to be performed depends
on the value in the ctype field of the GUE header. on the value in the ctype field of the GUE header.
5.5. Router and switch operation 5.5. Router and switch operation
Routers and switches SHOULD forward GUE packets as standard UDP/IP Routers and switches MUST forward GUE packets as standard UDP/IP
packets. The outer five-tuple should contain sufficient information packets.
<!-- CEP: Another puzzling unenforceable mandate. Why would
inclusion of GUE suddenly invalidate previous standards? -->
The outer five-tuple should contain sufficient information
to perform flow classification corresponding to the flow of the inner to perform flow classification corresponding to the flow of the inner
packet. A router does not normally need to parse a GUE header, and packet. A router does not normally need to parse a GUE header, and
none of the flags or extension fields in the GUE header are expected none of the flags or extension fields in the GUE header are expected
to affect routing. In cases where the outer five-tuple does not to affect routing. In cases where the outer five-tuple does not
provide sufficient entropy for flow classification, for instance UDP provide sufficient entropy for flow classification, for instance UDP
ports are fixed to provide connection semantics (section 5.6.1), then ports are fixed to provide connection semantics (section 5.6.1), then
the encapsulated packet MAY be parsed to determine flow entropy. the encapsulated packet MAY be parsed to determine flow entropy.
A router MUST NOT modify a GUE header when forwarding a packet. It A router MUST NOT modify a GUE header when forwarding a packet. It
MAY encapsulate a GUE packet in another GUE packet, for instance to MAY encapsulate a GUE packet in another GUE packet, for instance to
implement a network tunnel (i.e. by encapsulating an IP packet with a implement a network tunnel (i.e. by encapsulating an IP packet with a
GUE payload in another IP packet as a GUE payload). In this case, the GUE payload in another IP packet as a GUE payload). In this case, the
router takes the role of an encapsulator, and the corresponding router takes the role of an encapsulator, and the corresponding
decapsulator is the logical endpoint of the tunnel. When decapsulator is the logical endpoint of the tunnel. When
encapsulating a GUE packet within another GUE packet, there are no encapsulating a GUE packet within another GUE packet, there are no
provisions to automatically copy flags or fields to the outer GUE provisions to automatically copy flags or fields to the outer GUE
header. Each layer of encapsulation is considered independent. header. Each layer of encapsulation is considered independent.
<!-- CEP: Does this document intend to update RFC 2983, 6040? -->
5.6. Middlebox interactions 5.6. Middlebox interactions
A middlebox MAY interpret some flags and extension fields of the GUE A middlebox MAY interpret some flags and extension fields of the GUE
header for classification purposes, but is not required to understand header for classification purposes, but is not required to understand
any of the flags or extension fields in GUE packets. A middlebox MUST any of the flags or extension fields in GUE packets. A middlebox MUST
NOT drop a GUE packet merely because there are flags unknown to it. NOT drop a GUE packet merely because there are flags unknown to it.
<!-- CEP: unenforceable. -->
The header length in the GUE header allows a middlebox to inspect the The header length in the GUE header allows a middlebox to inspect the
payload packet without needing to parse the flags or extension payload packet without needing to parse the flags or extension
fields. fields.
5.6.1. Inferring connection semantics 5.6.1. Inferring connection semantics
A middlebox might infer bidirectional connection semantics for a UDP A middlebox might infer bidirectional connection semantics for a UDP
flow. For instance, a stateful firewall might create a five-tuple flow. For instance, a stateful firewall might create a five-tuple
rule to match flows on egress, and a corresponding five-tuple rule rule to match flows on egress, and a corresponding five-tuple rule
for matching ingress packets where the roles of source and for matching ingress packets where the roles of source and
skipping to change at page 20, line 34 skipping to change at page 20, line 36
To operate in this environment, a GUE tunnel should be configured to To operate in this environment, a GUE tunnel should be configured to
assume connected semantics defined by the UDP five tuple and the use assume connected semantics defined by the UDP five tuple and the use
of GUE encapsulation needs to be symmetric between both endpoints. of GUE encapsulation needs to be symmetric between both endpoints.
The source port set in the UDP header MUST be the destination port The source port set in the UDP header MUST be the destination port
the peer would set for replies. In this case, the UDP source port for the peer would set for replies. In this case, the UDP source port for
a tunnel would be a fixed value and not set to be flow entropy as a tunnel would be a fixed value and not set to be flow entropy as
described in section 5.11. described in section 5.11.
The selection of whether to make the UDP source port fixed or set to The selection of whether to make the UDP source port fixed or set to
a flow entropy value for each packet sent SHOULD be configurable for a flow entropy value for each packet sent SHOULD be configurable for
a tunnel. The default MUST be to set the flow entropy value in the a tunnel.
<!-- CEP: The SHOULD is a mandate on the configuration of GUE, not
on GUE protocol. -->
The default MUST be to set the flow entropy value in the
UDP source port. UDP source port.
5.6.2. NAT 5.6.2. NAT
IP address and port translation can be performed on the UDP/IP IP address and port translation can be performed on the UDP/IP
headers adhering to the requirements for NAT with UDP [RFC4787]. In headers adhering to the requirements for NAT with UDP [RFC4787]. In
the case of stateful NAT, connection semantics MUST be applied to a the case of stateful NAT, connection semantics MUST be applied to a
GUE tunnel as described in section 5.6.1. GUE endpoints MAY also GUE tunnel as described in section 5.6.1. GUE endpoints MAY also
invoke STUN [RFC5389] or ICE [RFC5245] to manage NAT port mappings invoke STUN [RFC5389] or ICE [RFC5245] to manage NAT port mappings
for encapsulations. for encapsulations.
5.7. Checksum Handling 5.7. Checksum Handling
The potential for mis-delivery of packets due to corruption of IP, The potential for mis-delivery of packets due to corruption of IP,
UDP, or GUE headers needs to be considered. Historically, the UDP UDP, or GUE headers needs to be considered. Historically, the UDP
checksum would be considered sufficient as a check against corruption checksum would be considered sufficient as a check against corruption
of either the UDP header and payload or the IP addresses. of either the UDP header and payload or the IP addresses.
Encapsulation protocols, such as GUE, can be originated or terminated Encapsulation protocols, such as GUE, can be originated or terminated
on devices incapable of computing the UDP checksum for packet. This on devices incapable of computing the UDP checksum for packet.
<!-- CEP: citation needed for such devices. -->
This
section discusses the requirements around checksum and alternatives section discusses the requirements around checksum and alternatives
that might be used when an endpoint does not support UDP checksum. that might be used when an endpoint does not support UDP checksum.
5.7.1. Requirements 5.7.1. Requirements
One of the following requirements MUST be met: One of the following requirements MUST be met:
<!-- CEP: This is a tautology. Either the checksum is zero, or not. -->
o UDP checksums are enabled (for IPv4 or IPv6). o UDP checksums are enabled (for IPv4 or IPv6).
<!-- CEP: Is GUE defined in this document for anything other
than IPv[4,6]? Is UDP defined for anything else?? -->
o The GUE header checksum is used (defined in [GUEEXTEN]). o The GUE header checksum is used (defined in [GUEEXTEN]).
o Use zero UDP checksums. This is always permissible with IPv4; in o Use zero UDP checksums. This is always permissible with IPv4; in
IPv6, they can only be used in accordance with applicable IPv6, they can only be used in accordance with applicable
requirements in [RFC8086], [RFC6935], and [RFC6936]. requirements in [RFC8086], [RFC6935], and [RFC6936].
5.7.2. UDP Checksum with IPv4 5.7.2. UDP Checksum with IPv4
For UDP in IPv4, the UDP checksum MUST be processed as specified in For UDP in IPv4, the UDP checksum MUST be processed as specified in
[RFC768] and [RFC1122] for both transmit and receive. An [RFC768] and [RFC1122] for both transmit and receive. An
encapsulator MAY set the UDP checksum to zero for performance or encapsulator MAY set the UDP checksum to zero for performance or
implementation considerations. The IPv4 header includes a checksum implementation considerations. The IPv4 header includes a checksum
that protects against mis-delivery of the packet due to corruption that protects against mis-delivery of the packet due to corruption
of IP addresses. The UDP checksum potentially provides protection of IP addresses. The UDP checksum potentially provides protection
against corruption of the UDP header, GUE header, and GUE payload. against corruption of the UDP header, GUE header, and GUE payload.
Enabling or disabling the use of checksums is a deployment Enabling or disabling the use of checksums is a deployment
consideration that should take into account the risk and effects of consideration that SHOULD take into account the risk and effects of
packet corruption, and whether the packets in the network are packet corruption, and whether the packets in the network are
already adequately protected by other, possibly stronger mechanisms, already adequately protected by other, possibly stronger mechanisms,
such as the Ethernet CRC. If an encapsulator sets a zero UDP such as the Ethernet CRC. If an encapsulator sets a zero UDP
checksum for IPv4, it SHOULD use the GUE header checksum as checksum for IPv4, it SHOULD use the GUE header checksum as
described in [GUEEXTEN] assuming there are no other mechanisms used described in [GUEEXTEN] assuming there are no other mechanisms used
to protect the GUE packet. to protect the GUE packet.
<!-- CEP: This last sentence seems to disregard the previous sentence,
which would then obviate the need for a GUE checksum. -->
When a decapsulator receives a packet, the UDP checksum field MUST When a decapsulator receives a packet, the UDP checksum field MUST
be processed. If the UDP checksum is non-zero, the decapsulator MUST be processed. If the UDP checksum is non-zero, the decapsulator MUST
verify the checksum before accepting the packet. By default, a verify the checksum before accepting the packet. By default, a
decapsulator SHOULD accept UDP packets with a zero checksum. A node decapsulator SHOULD accept UDP packets with a zero checksum. A node
MAY be configured to disallow zero checksums per [RFC1122]. MAY be configured to disallow zero checksums per [RFC1122].
Configuration of zero checksums can be selective. For instance, zero Configuration of zero checksums can be selective. For instance, zero
checksums might be disallowed from certain hosts that are known to checksums might be disallowed from certain hosts that are known to
be traversing paths subject to packet corruption. If verification of be traversing paths subject to packet corruption. If verification of
a non-zero checksum fails, a decapsulator lacks the capability to a non-zero checksum fails, a decapsulator lacks the capability to
skipping to change at page 22, line 14 skipping to change at page 22, line 14
5.7.3. UDP Checksum with IPv6 5.7.3. UDP Checksum with IPv6
In IPv6, there is no checksum in the IPv6 header that protects In IPv6, there is no checksum in the IPv6 header that protects
against mis-delivery due to address corruption. Therefore, when GUE against mis-delivery due to address corruption. Therefore, when GUE
is used over IPv6, either the UDP checksum or the GUE header is used over IPv6, either the UDP checksum or the GUE header
checksum SHOULD be used unless there are alternative mechanisms in checksum SHOULD be used unless there are alternative mechanisms in
use that protect against misdelivery. The UDP checksum and GUE use that protect against misdelivery. The UDP checksum and GUE
header checksum SHOULD NOT be used at the same time since that would header checksum SHOULD NOT be used at the same time since that would
be mostly redundant. be mostly redundant.
<!-- CEP: Why not simply mandate the UDP checksum? -->
If neither the UDP checksum or the GUE header checksum is used, then If neither the UDP checksum or the GUE header checksum is used, then
the requirements for using zero IPv6 UDP checksums in [RFC6935] and the requirements for using zero IPv6 UDP checksums in [RFC6935] and
[RFC6936] MUST be met. [RFC6936] MUST be met.
When a decapsulator receives a packet, the UDP checksum field MUST When a decapsulator receives a packet, the UDP checksum field MUST
be processed. If the UDP checksum is non-zero, the decapsulator MUST be processed. If the UDP checksum is non-zero, the decapsulator MUST
verify the checksum before accepting the packet. By default a verify the checksum before accepting the packet. By default a
decapsulator MUST only accept UDP packets with a zero checksum if decapsulator MUST only accept UDP packets with a zero checksum if
the GUE header checksum is used and is verified. If verification of the GUE header checksum is used and is verified. If verification of
a non-zero checksum fails, a decapsulator lacks the capability to a non-zero checksum fails, a decapsulator lacks the capability to
verify a non-zero checksum, or a packet with a zero-checksum and no verify a non-zero checksum, or a packet with a zero-checksum and no
GUE header checksum was received, the packet MUST be dropped. GUE header checksum was received, the packet MUST be dropped.
5.8. MTU and fragmentation 5.8. MTU and fragmentation
Standard conventions for handling of MTU (Maximum Transmission Unit) Standard conventions for handling of MTU (Maximum Transmission Unit)
and fragmentation in conjunction with networking tunnels and fragmentation in conjunction with networking tunnels
(encapsulation of layer 2 or layer 3 packets) SHOULD be followed. (encapsulation of layer 2 or layer 3 packets) MUST be followed.
Details are described in MTU and Fragmentation Issues with In-the- Details are described in MTU and Fragmentation Issues with In-the-
Network Tunneling [RFC4459]. Network Tunneling [RFC4459].
If a packet is fragmented before encapsulation in GUE, all the If a packet is fragmented before encapsulation in GUE, all the
related fragments MUST be encapsulated using the same UDP source related fragments MUST be encapsulated using the same UDP source
port. An operator SHOULD set MTU to account for encapsulation port. An operator SHOULD set MTU to account for encapsulation
overhead and reduce the likelihood of fragmentation. overhead and reduce the likelihood of fragmentation.
Alternative to IP fragmentation, the GUE fragmentation extension can Alternative to IP fragmentation, the GUE fragmentation extension can
be used. GUE fragmentation is described in [GUEEXTEN]. be used. GUE fragmentation is described in [GUEEXTEN].
<!-- CEP: GUE fragmentation has to be specified in this document. -->
5.9. Congestion control 5.9. Congestion control
Per requirements of [RFC5405], if the IP traffic encapsulated with Per requirements of [RFC5405], if the IP traffic encapsulated with
GUE implements proper congestion control no additional mechanisms GUE implements proper congestion control no additional mechanisms
should be required. are required.
In the case that the encapsulated traffic does not implement any or In the case that the encapsulated traffic does not implement any or
sufficient control, or it is not known whether a transmitter will sufficient control, or it is not known whether a transmitter will
consistently implement proper congestion control, then congestion consistently implement proper congestion control, then congestion
control at the encapsulation layer MUST be provided per [RFC5405]. control at the encapsulation layer MUST be provided per [RFC5405].
Note that this case applies to a significant use case in network This applies to a significant use case in network
virtualization in which guests run third party networking stacks virtualization in which guests run third party networking stacks
that cannot be implicitly trusted to implement conformant congestion that cannot be implicitly trusted to implement conformant congestion
control. control.
Out of band mechanisms such as rate limiting, Managed Circuit Out of band mechanisms such as rate limiting, Managed Circuit
Breaker [RFC8084], or traffic isolation MAY be used to provide Breaker [RFC8084], or traffic isolation MAY be used to provide
rudimentary congestion control. For finer-grained congestion control rudimentary congestion control. For finer-grained congestion control
that allows alternate congestion control algorithms, reaction time that allows alternate congestion control algorithms, reaction time
within an RTT, and interaction with ECN, in-band mechanisms might be within an RTT, and interaction with ECN, in-band mechanisms might be
warranted. warranted.
skipping to change at page 23, line 42 skipping to change at page 23, line 42
independent of the encapsulation and is otherwise outside the scope independent of the encapsulation and is otherwise outside the scope
of this document. of this document.
5.11. Flow entropy for ECMP 5.11. Flow entropy for ECMP
5.11.1. Flow classification 5.11.1. Flow classification
A major objective of using GUE is that a network device can perform A major objective of using GUE is that a network device can perform
flow classification corresponding to the flow of the inner flow classification corresponding to the flow of the inner
encapsulated packet based on the contents in the outer headers. encapsulated packet based on the contents in the outer headers.
<!-- CEP: This sentence belongs in the Introduction. -->
Hardware devices commonly perform hash computations on packet Hardware devices commonly perform hash computations on packet
headers to classify packets into flows or flow buckets. Flow headers to classify packets into flows or flow buckets. Flow
classification is done to support load balancing of flows across a classification is done to support load balancing of flows across a
set of networking resources. Examples of such load balancing set of networking resources. Examples of such load balancing
techniques are Equal Cost Multipath routing (ECMP), port selection techniques are Equal Cost Multipath routing (ECMP), port selection
in Link Aggregation, and NIC device Receive Side Scaling (RSS). in Link Aggregation, and NIC device Receive Side Scaling (RSS).
<!-- CEP: citations needed. -->
Hashes are usually either a three-tuple hash of IP protocol, source Hashes are usually either a three-tuple hash of IP protocol, source
address, and destination address; or a five-tuple hash consisting of address, and destination address; or a five-tuple hash consisting of
IP protocol, source address, destination address, source port, and IP protocol, source address, destination address, source port, and
destination port. Typically, networking hardware will compute five- destination port. Typically, networking hardware will compute five-
tuple hashes for TCP and UDP, but only three-tuple hashes for other tuple hashes for TCP and UDP, but only three-tuple hashes for other
IP protocols. Since the five-tuple hash provides more granularity, IP protocols. Since the five-tuple hash provides more granularity,
load balancing can be finer-grained with better distribution. When a load balancing can be finer-grained with better distribution. When a
packet is encapsulated with GUE and connection semantics are not packet is encapsulated with GUE and connection semantics are not
applied, the source port in the outer UDP packet is set to a flow applied, the source port in the outer UDP packet is set to a flow
entropy value that corresponds to the flow of the inner packet. When entropy value that corresponds to the flow of the inner packet. When
a device computes a five-tuple hash on the outer UDP/IP header of a a device computes a five-tuple hash on the outer UDP/IP header of a
GUE packet, the resultant value classifies the packet per its inner GUE packet, the resultant value classifies the packet per its inner
flow. flow.
Examples of deriving flow entropy for encapsulation are: Examples of deriving flow entropy for encapsulation are:
o If the encapsulated packet is a layer 4 packet, TCP/IPv4 for o If the encapsulated packet is a layer 4 packet, TCPv4 for
instance, the flow entropy could be based on the canonical five- instance, the flow entropy could be based on the canonical five-
tuple hash of the inner packet. tuple hash of the inner packet.
o If the encapsulated packet is an AH transport mode packet with o If the encapsulated packet is an AH transport mode packet with
TCP as next header, the flow entropy could be a hash over a TCP as next header, the flow entropy could be a hash over a
three-tuple: TCP protocol and TCP ports of the encapsulated three-tuple: TCP protocol and TCP ports of the encapsulated
packet. packet.
o If a node is encrypting a packet using ESP tunnel mode and GUE o If a node is encrypting a packet using ESP tunnel mode and GUE
encapsulation, the flow entropy could be based on the contents encapsulation, the flow entropy could be based on the contents
of the clear-text packet. For instance, a canonical five-tuple of the clear-text packet. For instance, a canonical five-tuple
hash for a TCP/IP packet could be used. hash for a TCP/IP packet could be used.
[RFC6438] discusses methods to compute and set flow entropy value for [RFC6438] discusses methods to compute and set flow entropy value for
IPv6 flow labels. Such methods can also be used to create flow IPv6 flow labels. Such methods can also be used to create flow
entropy values for GUE. entropy values for GUE.
<!-- CEP: for interoperability, maybe more specifics are needed. -->
5.11.2. Flow entropy properties 5.11.2. Flow entropy properties
The flow entropy is the value set in the UDP source port of a GUE The flow entropy is the value set in the UDP source port of a GUE
packet. Flow entropy in the UDP source port SHOULD adhere to the packet. Flow entropy in the UDP source port SHOULD adhere to the
following properties: following properties:
o The value set in the source port is within the ephemeral port o The value set in the source port is within the ephemeral port
range (49152 to 65535 [RFC6335]). Since the high order two bits range (49152 to 65535 [RFC6335]). Since the high order two bits
of the port are set to one, this provides fourteen bits of of the port are set to one, this provides fourteen bits of
entropy for the value. entropy for the value.
o The flow entropy has a uniform distribution across encapsulated o The flow entropy has a uniform distribution across encapsulated
flows. flows.
o An encapsulator MAY occasionally change the flow entropy used o An encapsulator MAY occasionally change the flow entropy used
for an inner flow per its discretion (for security, route for an inner flow (for security, route
selection, etc). To avoid thrashing or flapping the value, the selection, etc). To avoid thrashing or flapping the value, the
flow entropy used for a flow SHOULD NOT change more than once flow entropy used for a flow SHOULD NOT change more than once
every thirty seconds (or a configurable value). every thirty seconds (or a configurable value).
<!-- CEP: This configurable value needs a name. -->
o Decapsulators, or any networking devices, SHOULD NOT attempt to o Decapsulators, or any networking devices, SHOULD NOT attempt to
interpret flow entropy as anything more than an opaque value. interpret flow entropy as anything more than an opaque value.
Neither should they attempt to reproduce the hash calculation Neither should they attempt to reproduce the hash calculation
used by an encapasulator in creating a flow entropy value. They used by an encapasulator in creating a flow entropy value. They
MAY use the value to match further receive packets for steering MAY use the value to match further receive packets for steering
decisions, but MUST NOT assume that the hash uniquely or decisions, but MUST NOT assume that the hash uniquely or
permanently identifies a flow. permanently identifies a flow.
o Input to the flow entropy calculation is not restricted to ports o Input to the flow entropy calculation is not restricted to ports
and addresses; input could include flow label from an IPv6 and addresses; input could include flow label from an IPv6
packet, SPI from an ESP packet, or other flow related state in packet, SPI from an ESP packet, or other flow related state in
the encapsulator that is not necessarily conveyed in the packet. the encapsulator that is not necessarily conveyed in the packet.
o The assignment function for flow entropy SHOULD be randomly o The assignment function for flow entropy SHOULD be randomly
seeded to mitigate denial of service attacks. The seed SHOULD be seeded to mitigate denial of service attacks. The seed SHOULD be
changed periodically. changed periodically.
<!-- CEP: This needs to be specified. -->
5.12 Negotiation of acceptable flags and extension fields 5.12 Negotiation of acceptable flags and extension fields
An encapsulator and decapsulator need to achieve agreement about GUE An encapsulator and decapsulator need to achieve agreement about GUE
parameters that will be used in communications. Parameters include parameters that will be used in communications. Parameters include
supported GUE variants, flags and extension fields that can be used, supported GUE variants, flags and extension fields that can be used,
security algorithms and keys, supported protocols and control security algorithms and keys, supported protocols and control
messages, etc. This document proposes different general methods to messages, etc. This document proposes different general methods to
accomplish this, however the details of implementing these are accomplish this, however the details of implementing these are
considered out of scope. considered out of scope.
skipping to change at page 26, line 9 skipping to change at page 26, line 9
needed for network virtualization). needed for network virtualization).
o Via security negotiation. Use of security typically implies a o Via security negotiation. Use of security typically implies a
key exchange between endpoints. Other GUE parameters may be key exchange between endpoints. Other GUE parameters may be
conveyed as part of that process. conveyed as part of that process.
6. Motivation for GUE 6. Motivation for GUE
This section presents the motivation for GUE with respect to other This section presents the motivation for GUE with respect to other
encapsulation methods. encapsulation methods.
<!-- CEP: This section needs to be moved much closer to the beginning
of the document, perhaps just before Terminology. -->
6.1. Benefits of GUE 6.1. Benefits of GUE
* GUE is a generic encapsulation protocol. GUE can encapsulate * GUE is a generic encapsulation protocol. GUE can encapsulate
protocols that are represented by an IP protocol number. This protocols that are represented by an IP protocol number.
includes layer 2, layer 3, and layer 4 protocols.
* GUE is an extensible encapsulation protocol. Standardized * GUE is an extensible encapsulation protocol. Standardized
optional data such as security, virtual networking identifiers, optional data such as security, virtual networking identifiers,
fragmentation are being defined. fragmentation are being defined.
<!-- CEP: citations requested. -->
* For extensilbity, GUE uses flag fields as opposed to TLVs as * For extensibility, GUE uses flag fields as opposed to TLVs as
some other encapsulation protocols do. Flag fields are strictly some other encapsulation protocols do. Flag fields are strictly
ordered, allow random access, and are efficient in use of header ordered, allow random access, and are efficient in use of header
space. space.
* GUE allows private data to be sent as part of the encapsulation. * GUE allows private data to be sent as part of the encapsulation.
This permits experimentation or customization in deployment. This permits experimentation or customization in deployment.
* GUE allows sending of control messages such as OAM using the * GUE allows sending of control messages such as OAM using the
same GUE header format (for routing purposes) as normal data same GUE header format (for routing purposes) as normal data
messages. messages.
* GUE maximizes deliverability of non-UDP and non-TCP protocols. * GUE maximizes deliverability of non-UDP and non-TCP protocols.
<!-- CEP: This claim relies on the assumption the intermediate
routing points are happy with random UDP payloads. -->
* GUE provides a means for exposing per flow entropy for ECMP for * GUE provides a means for exposing per flow entropy for ECMP for
atypical protocols such as SCTP, DCCP, ESP, etc. atypical protocols such as SCTP, DCCP, ESP, etc.
6.2 Comparison of GUE to other encapsulations 6.2 Comparison of GUE to other encapsulations
<!-- CEP: This section also belongs much earlier in the document. -->
A number of different encapsulation techniques have been proposed for A number of different encapsulation techniques have been proposed for
the encapsulation of one protocol over another. EtherIP [RFC3378] the encapsulation of one protocol over another. EtherIP [RFC3378]
provides layer 2 tunneling of Ethernet frames over IP. GRE [RFC2784], provides tunneling of Ethernet frames over IP. GRE [RFC2784],
MPLS [RFC4023], and L2TP [RFC2661] provide methods for tunneling MPLS [RFC4023], and L2TP [RFC2661] provide methods for tunneling
layer 2 and layer 3 packets over IP. NVGRE [RFC7637] and VXLAN layer 2 and layer 3 packets over IP. NVGRE [RFC7637] and VXLAN
[RFC7348] are proposals for encapsulation of layer 2 packets for [RFC7348] are proposals for encapsulation of layer 2 packets for
network virtualization. IPIP [RFC2003] and Generic packet tunneling network virtualization. IPIP [RFC2003] and Generic packet tunneling
in IPv6 [RFC2473] provide methods for tunneling IP packets over IP. in IPv6 [RFC2473] provide methods for tunneling IP packets over IP.
Several proposals exist for encapsulating packets over UDP including Several proposals exist for encapsulating packets over UDP including
ESP over UDP [RFC3948], TCP directly over UDP [TCPUDP], VXLAN ESP over UDP [RFC3948], TCP directly over UDP [TCPUDP], VXLAN
[RFC7348], LISP [RFC6830] which encapsulates layer 3 packets, [RFC7348], LISP [RFC6830] which encapsulates layer 3 packets,
MPLS/UDP [RFC7510], GENEVE [GENEVE], and GRE-in-UDP Encapsulation MPLS/UDP [RFC7510], GENEVE [GENEVE], and GRE-in-UDP Encapsulation
[RFC8086]. [RFC8086].
GUE has the following discriminating features: GUE has the following discriminating features:
o UDP encapsulation leverages specialized network device o UDP encapsulation leverages specialized network device
processing for efficient transport. The semantics for using the processing for efficient transport. The semantics for using the
UDP source port for flow entropy as input to ECMP are defined in UDP source port for flow entropy as input to ECMP are defined in
section 5.11. section 5.11.
o GUE permits encapsulation of arbitrary IP protocols, which o GUE permits encapsulation of arbitrary IP protocols.
includes layer 2 3, and 4 protocols.
o Multiple protocols can be multiplexed over a single UDP port o Multiple protocols can be multiplexed over a single UDP port
number. This is in contrast to techniques to encapsulate number. This is in contrast to techniques to encapsulate
protocols over UDP using a protocol specific port number (such protocols over UDP using a protocol specific port number (such
as ESP/UDP, GRE/UDP, SCTP/UDP). GUE provides a uniform and as ESP/UDP, GRE/UDP, SCTP/UDP). GUE provides a uniform and
extensible mechanism for encapsulating all IP protocols in UDP extensible mechanism for encapsulating all IP protocols in UDP
with minimal overhead (four bytes of additional header). with minimal overhead (as few as four bytes of additional header).
o GUE is extensible. New flags and extension fields can be o GUE is extensible. New flags and extension fields can be
defined. defined.
o The GUE header includes a header length field. This allows a o The GUE header includes a header length field. This allows a
network node to inspect an encapsulated packet without needing network node to inspect an encapsulated packet without needing
to parse the full encapsulation header. to parse the full encapsulation header.
o Private data in the encapsulation header allows local o Private data in the encapsulation header allows local
customization and experimentation while being compatible with customization and experimentation while being compatible with
processing in network nodes (routers and middleboxes). processing in network nodes (routers and middleboxes).
o GUE includes both data messages (encapsulation of packets) and o GUE includes both data messages (encapsulation of packets) and
control messages (such as OAM). control messages (such as OAM).
o The flags-field model facilitates efficient implementation of o The flags-field model facilitates efficient implementation of
extensibility in hardware. For instance, a TCAM can be use to extensibility in hardware. For instance, a TCAM can be used to
parse a known set of N flags where the number of entries in the parse a known set of N flags where the number of entries in the
TCAM is 2^N. By comparison, the number of TCAM entries needed to TCAM is 2^N. By comparison, the number of TCAM entries needed to
parse a set of N arbitrarily ordered TLVS is approximately e*N!. parse a set of N arbitrarily ordered TLVS is approximately e*N!.
o GUE includes a variant that encapsulates IPv4 and IPv6 packets o GUE includes a variant that encapsulates IPv4 and IPv6 packets
directly within UDP. directly within UDP.
7. Security Considerations 7. Security Considerations
There are two important considerations of security with respect to There are two important considerations of security with respect to
GUE. GUE.
o Authentication and integrity of the GUE header. o Authentication and integrity of the GUE header.
o Authentication, integrity, and confidentiality of the GUE o Authentication, integrity, and confidentiality of the GUE
payload. payload.
GUE security is provided by extensions for security defined in GUE security is provided by extensions for security defined in
[GUEEXTEN]. These extensions include methods to authenticate the GUE [GUEEXTEN]. These extensions include methods to authenticate the GUE
header and encrypt the GUE payload. header and encrypt the GUE payload.
<!-- CEP: Please explain why GUE requires yet another mechanism. -->
The GUE header can be authenticated using a security extension for an The GUE header can be authenticated using a security extension for an
HMAC. Securing the GUE payload can be accomplished use of the GUE HMAC. Securing the GUE payload can be accomplished use of the GUE
Payload Transform. This extension can be used to perform DTLS in the Payload Transform. This extension can be used to perform DTLS in the
payload of a GUE packet to encrypt the payload. payload of a GUE packet to encrypt the payload.
A hash function for computing flow entropy (section 5.11) SHOULD be A hash function for computing flow entropy (section 5.11) SHOULD be
randomly seeded to mitigate some possible denial service attacks. randomly seeded to mitigate some possible denial service attacks.
<!-- CEP: Citation needed for "possible" and for how a random seed
deters them. -->
8. IANA Considerations 8. IANA Considerations
8.1. UDP source port 8.1. UDP source port
A user UDP port number assignment for GUE has been assigned: A user UDP port number assignment for GUE has been assigned:
Service Name: gue Service Name: gue
Transport Protocol(s): UDP Transport Protocol(s): UDP
Assignee: Tom Herbert <tom@herbertland.com> Assignee: Tom Herbert <tom@herbertland.com>
skipping to change at page 33, line 30 skipping to change at page 33, line 30
Generic Network Virtualization Encapsulation", draft-ietf- Generic Network Virtualization Encapsulation", draft-ietf-
nvo3-geneve-05 nvo3-geneve-05
[LCO] Cree, E., https://www.kernel.org/doc/Documentation/ [LCO] Cree, E., https://www.kernel.org/doc/Documentation/
networking/checksum-offloads.txt networking/checksum-offloads.txt
Appendix A: NIC processing for GUE Appendix A: NIC processing for GUE
This appendix provides some guidelines for Network Interface Cards This appendix provides some guidelines for Network Interface Cards
(NICs) to implement common offloads and accelerations to support GUE. (NICs) to implement common offloads and accelerations to support GUE.
Note that most of this discussion is generally applicable to other Most of this discussion is generally applicable to other
methods of UDP based encapsulation. methods of UDP based encapsulation.
<!-- CEP: The last sentence means these discussions do not belong here.-->
A.1. Receive multi-queue A.1. Receive multi-queue
Contemporary NICs support multiple receive descriptor queues (multi- Contemporary NICs support multiple receive descriptor queues (multi-
queue). Multi-queue enables load balancing of network processing for queue). Multi-queue enables load balancing of network processing for
a NIC across multiple CPUs. On packet reception, a NIC selects the a NIC across multiple CPUs. On packet reception, a NIC selects the
appropriate queue for host processing. Receive Side Scaling is a appropriate queue for host processing. Receive Side Scaling is a
common method which uses the flow hash for a packet to index an common method which uses the flow hash for a packet to index an
indirection table where each entry stores a queue number. Flow indirection table where each entry stores a queue number. Flow
Director and Accelerated Receive Flow Steering (aRFS) allow a host to Director and Accelerated Receive Flow Steering (aRFS) allow a host to
program the queue that is used for a given flow which is identified program the queue that is used for a given flow which is identified
either by an explicit five-tuple or by the flow's hash. either by an explicit five-tuple or by the flow's hash.
<!-- CEP: citations needed for RSS, aRFS, etc. -->
GUE encapsulation is compatible with multi-queue NICs that support GUE encapsulation is compatible with multi-queue NICs that support
five-tuple hash calculation for UDP/IP packets as input to RSS. The five-tuple hash calculation for UDP/IP packets as input to RSS. The
flow entropy in the UDP source port ensures classification of the flow entropy in the UDP source port ensures classification of the
encapsulated flow even in the case that the outer source and encapsulated flow even in the case that the outer source and
destination addresses are the same for all flows (e.g. all flows are destination addresses are the same for all flows (e.g. all flows are
going over a single tunnel). going over a single tunnel).
By default, UDP RSS support is often disabled in NICs to avoid out- By default, UDP RSS support is often disabled in NICs to avoid out-
of-order reception that can occur when UDP packets are fragmented. As of-order reception that can occur when UDP packets are fragmented. As
<!-- CEP: citation needed for this claim. -->
<!-- CEP: should explain why fragmented packets cause reordering. -->
discussed above, fragmentation of GUE packets is mostly avoided by discussed above, fragmentation of GUE packets is mostly avoided by
fragmenting packets before entering a tunnel, GUE fragmentation, path fragmenting packets before entering a tunnel, GUE fragmentation, path
<!-- CEP: This makes the case for specifying GUE fragmentation as part
of this document. -->
MTU discovery in higher layer protocols, or operator adjusting MTUs. MTU discovery in higher layer protocols, or operator adjusting MTUs.
Other UDP traffic might not implement such procedures to avoid Other UDP traffic might not implement such procedures to avoid
fragmentation, so enabling UDP RSS support in the NIC might be a fragmentation, so enabling UDP RSS support in the NIC might be a
considered tradeoff during configuration. considered tradeoff during configuration.
A.2. Checksum offload A.2. Checksum offload
Many NICs provide capabilities to calculate standard ones complement Many NICs provide capabilities to calculate standard ones complement
payload checksum for packets in transmit or receive. When using GUE payload checksum for packets in transmit or receive. When using GUE
encapsulation, there are at least two checksums that are of interest: encapsulation, there are at least two checksums that are of interest:
skipping to change at page 34, line 28 skipping to change at page 34, line 32
the outer header. the outer header.
A.2.1. Transmit checksum offload A.2.1. Transmit checksum offload
NICs can provide a protocol agnostic method to offload transmit NICs can provide a protocol agnostic method to offload transmit
checksum (NETIF_F_HW_CSUM in Linux parlance) that can be used with checksum (NETIF_F_HW_CSUM in Linux parlance) that can be used with
GUE. In this method, the host provides checksum related parameters in GUE. In this method, the host provides checksum related parameters in
a transmit descriptor for a packet. These parameters include the a transmit descriptor for a packet. These parameters include the
starting offset of data to checksum, the length of data to checksum, starting offset of data to checksum, the length of data to checksum,
and the offset in the packet where the computed checksum is to be and the offset in the packet where the computed checksum is to be
written. The host initializes the checksum field to pseudo header written. The host initializes the checksum field to a pseudo header
checksum. checksum.
In the case of GUE, the checksum for an encapsulated transport layer In the case of GUE, the checksum for an encapsulated transport layer
packet, a TCP packet for instance, can be offloaded by setting the packet, a TCP packet for instance, can be offloaded by setting the
appropriate checksum parameters. appropriate checksum parameters.
NICs typically can offload only one transmit checksum per packet, so NICs typically can offload only one transmit checksum per packet, so
simultaneously offloading both an inner transport packet's checksum simultaneously offloading both an inner transport packet's checksum
and the outer UDP checksum is likely not possible. and the outer UDP checksum is likely not possible.
skipping to change at page 35, line 40 skipping to change at page 35, line 40
for the outer UDP header in an encapsulation, checksum conversion can for the outer UDP header in an encapsulation, checksum conversion can
be done so that the checksum-complete value is derived and can be be done so that the checksum-complete value is derived and can be
used by the stack to validate checksums in the encapsulated packet. used by the stack to validate checksums in the encapsulated packet.
A.3. Transmit Segmentation Offload A.3. Transmit Segmentation Offload
Transmit Segmentation Offload (TSO) is a NIC feature where a host Transmit Segmentation Offload (TSO) is a NIC feature where a host
provides a large (>MTU size) TCP packet to the NIC, which in turn provides a large (>MTU size) TCP packet to the NIC, which in turn
splits the packet into separate segments and transmits each one. This splits the packet into separate segments and transmits each one. This
is useful to reduce CPU load on the host. is useful to reduce CPU load on the host.
<!-- CEP: Citations needed! -->
The process of TSO can be generalized as: The process of TSO can be generalized as:
- Split the TCP payload into segments which allow packets with - Split the TCP payload into segments which allow packets with
size less than or equal to MTU. size less than or equal to MTU.
- For each created segment: - For each created segment:
1. Replicate the TCP header and all preceding headers of the 1. Replicate the TCP header and all preceding headers of the
original packet. original packet.
2. Set payload length fields in any headers to reflect the 2. Set payload length fields in any headers to reflect the
length of the segment. length of the segment.
3. Set TCP sequence number to correctly reflect the offset of 3. Set TCP sequence number to correctly reflect the offset of
the TCP data in the stream. the TCP data in the stream.
<!-- CEP: This might conflict with the TCP sequence number chosen by
the TCP layer actually creating the segments. -->
4. Recompute and set any checksums that either cover the payload 4. Recompute and set any checksums that either cover the payload
of the packet or cover header which was changed by setting a of the packet or cover header which was changed by setting a
payload length. payload length.
Following this general process, TSO can be extended to support TCP Following this general process, TSO can be extended to support TCP
encapsulation in GUE. For each segment the Ethernet, outer IP, UDP encapsulation in GUE. For each segment the Ethernet, outer IP, UDP
header, GUE header, inner IP header (if tunneling), and TCP headers header, GUE header, inner IP header (if tunneling), and TCP headers
are replicated. Any packet length header fields need to be set are replicated. Any packet length header fields need to be set
properly (including the length in the outer UDP header), and properly (including the length in the outer UDP header), and
checksums need to be set correctly (including the outer UDP checksum checksums need to be set correctly (including the outer UDP checksum
if being used). if being used).
To facilitate TSO with GUE, it is recommended that extension fields To facilitate TSO with GUE, it is recommended that extension fields
do not contain values that need to be updated on a per segment basis. do not contain values that need to be updated on a per segment basis.
For example, extension fields should not include checksums, lengths, For example, extension fields should not include checksums, lengths,
or sequence numbers that refer to the payload. If the GUE header does or sequence numbers that refer to the payload. If the GUE header does
not contain such fields then the TSO engine only needs to copy the not contain such fields then the TSO engine only needs to copy the
bits in the GUE header when creating each segment and does not need bits in the GUE header when creating each segment and does not need
to parse the GUE header. to parse the GUE header.
<!-- CEP: This makes appendix A.4 normative. And if it's normative, it
really needs to be in the main body of the text. -->
A.4. Large Receive Offload A.4. Large Receive Offload
Large Receive Offload (LRO) is a NIC feature where packets of a TCP Large Receive Offload (LRO) is a NIC feature where packets of a TCP
connection are reassembled, or coalesced, in the NIC and delivered to connection are reassembled, or coalesced, in the NIC and delivered to
the host as one large packet. This feature can reduce CPU utilization the host as one large packet. This feature can reduce CPU utilization
in the host. in the host.
<!-- CEP: Citations needed! -->
LRO requires significant protocol awareness to be implemented LRO requires significant protocol awareness to be implemented
correctly and is difficult to generalize. Packets in the same flow correctly and is difficult to generalize. Packets in the same flow
need to be unambiguously identified. In the presence of tunnels or need to be unambiguously identified. In the presence of tunnels or
network virtualization, this may require more than a five-tuple match network virtualization, this may require more than a five-tuple match
(for instance packets for flows in two different virtual networks may (for instance packets for flows in two different virtual networks may
have identical five-tuples). Additionally, a NIC needs to perform have identical five-tuples). Additionally, a NIC needs to perform
validation over packets that are being coalesced, and needs to validation over packets that are being coalesced, and needs to
fabricate a single meaningful header from all the coalesced packets. fabricate a single meaningful header from all the coalesced packets.
The conservative approach to supporting LRO for GUE would be to The conservative approach to supporting LRO for GUE would be to
assign packets to the same flow only if they have identical five- assign packets to the same flow only if they have identical five-
tuple and were encapsulated the same way. That is the outer IP tuple and were encapsulated the same way. That is the outer IP
addresses, the outer UDP ports, GUE protocol, GUE flags and fields, addresses, the outer UDP ports, GUE protocol, GUE flags and fields,
and inner five tuple are all identical. and inner five tuple are all identical.
<!-- CEP: sounds like this ought to be deleted. -->
Appendix B: Implementation considerations Appendix B: Implementation considerations
This appendix is informational and does not constitute a normative This appendix is informational and does not constitute a normative
part of this document. part of this document.
B.1. Priveleged ports B.1. Privileged ports
Using the source port to contain a flow entropy value disallows the Using the source port to contain a flow entropy value disallows the
security method of a receiver enforcing that the source port be a security method of a receiver enforcing that the source port be a
privileged port. Privileged ports are defined by some operating privileged port. Privileged ports are defined by some operating
systems to restrict source port binding. Unix, for instance, systems to restrict source port binding. Unix, for instance,
considered port number less than 1024 to be privileged. considered port number less than 1024 to be privileged.
Enforcing that packets are sent from a privileged port is widely Enforcing that packets are sent from a privileged port is widely
considered an inadequate security mechanism and has been mostly considered an inadequate security mechanism and has been mostly
deprecated. To approximate this behavior, an implementation could deprecated.
<!-- CEP: citation needed. -->
To approximate this behavior, an implementation could
restrict a user from sending a packet destined to the GUE port restrict a user from sending a packet destined to the GUE port
without proper credentials. without proper credentials.
<!-- CEP: this is not at all specific to GUE. -->
B.2. Setting flow entropy as a route selector B.2. Setting flow entropy as a route selector
An encapsulator generating flow entropy in the UDP source port could An encapsulator generating flow entropy in the outer UDP source port could
modulate the value to perform a type of multipath source routing. modulate the value to perform a type of multipath source routing.
Assuming that networking switches perform ECMP based on the flow Assuming that networking switches perform ECMP based on the flow
hash, a sender can affect the path by altering the flow entropy. For hash, a sender can affect the path by altering the flow entropy. For
instance, a host can store a flow hash in its protocol control block instance, a host can store a flow hash in its protocol control block
(PCB) for an inner flow, and might alter the value upon detecting (PCB) for an inner flow, and might alter the value upon detecting
that packets are traversing a lossy path. Changing the flow entropy that packets are traversing a lossy path. Changing the flow entropy
for a flow SHOULD be subject to hysteresis (at most once every thirty for a flow SHOULD be subject to hysteresis (at most once every thirty
seconds) to limit the number of out of order packets. seconds) to limit the number of out of order packets.
<!-- CEP: If the appendix is not normative, it cannot place mandates.-->
B.3. Hardware protocol implementation considerations B.3. Hardware protocol implementation considerations
Low level data path protocols, such is GUE, are often supported in Low level data path protocols, such as GUE, are often supported in
high speed network device hardware. Variable length header (VLH) high speed network device hardware. Variable length header (VLH)
protocols like GUE are often considered difficult to efficiently protocols like GUE are often considered difficult to efficiently
implement in hardware. In order to retain the important implement in hardware. In order to retain the important
characteristics of an extensible and robust protocol, hardware characteristics of an extensible and robust protocol, hardware
vendors may practice "constrained flexibility". In this model, only vendors may practice "constrained flexibility". In this model, only
certain combinations or protocol header parameterizations are certain combinations or protocol header parameterizations are
implemented in hardware fast path. Each such parameterization is implemented in hardware fast path. Each such parameterization is
fixed length so that the particular instance can be optimized as a fixed length so that the particular instance can be optimized as a
fixed length protocol. In the case of GUE this constitutes specific fixed length protocol. In the case of GUE this constitutes specific
combinations of GUE flags, fields, and next protocol. The selected combinations of GUE flags, fields, and next protocol. The selected
combinations would naturally be the most common cases which form the combinations would naturally be the most common cases which form the
"fast path", and other combinations are assumed to take the "slow "fast path", and other combinations are assumed to take the "slow
path". path".
In time, needs and requirements of the protocol may change which may In time, needs and requirements of the protocol may change which may
manifest themselves as new parameterizations to be supported in the manifest themselves as new parameterizations to be supported in the
fast path. To allow this extensibility, a device practicing fast path. To allow this extensibility, a device practicing
constrained flexibility should allow the fast path parameterizations constrained flexibility should allow the fast path parameterizations
to be programmable. to be programmable.
<!-- CEP: This appendix is very generic and could be deleted without
harm. -->
Authors' Addresses Authors' Addresses
Tom Herbert Tom Herbert
Quantonium Quantonium
4701 Patrick Henry 4701 Patrick Henry
Santa Clara, CA 95054 Santa Clara, CA 95054
US US
Email: tom@herbertland.com Email: tom@herbertland.com
Lucy Yong Lucy Yong
Huawei USA Huawei USA
<!-- CEP: Lucy Yong is no longer a Huawei employee. I suggest
moving her name to a Contributors section. -->
5340 Legacy Dr. 5340 Legacy Dr.
Plano, TX 75024 Plano, TX 75024
US US
Email: lucy.yong@huawei.com Email: lucy.yong@huawei.com
Osama Zia Osama Zia
Microsoft Microsoft
1 Microsoft Way 1 Microsoft Way
Redmond, WA 98029 Redmond, WA 98029
 End of changes. 114 change blocks. 
81 lines changed or deleted 199 lines changed or added

This html diff was produced by rfcdiff 1.47. The latest version is available from http://tools.ietf.org/tools/rfcdiff/