[pm-dir] Perf metrics: Fixed parameters/Traffic filter and hybrid type (draft-tgraf-opsawg-ipfix-on-path-telemetry)

[Benoit Claise <[e-mail]>]

Dear Performance metrics experts,

Thinking some more about draft-tgraf-opsawg-ipfix-on-path-telemetry, we 
have a question in connection to our draft 
draft-tgraf-opsawg-ipfix-on-path-telemetry-01.

See 
https://datatracker.ietf.org/doc/html/draft-tgraf-opsawg-ipfix-on-path-telemetry-01#name-fixed-parameters

    Traffic Filters:¶
    <https://datatracker.ietf.org/doc/html/draft-tgraf-opsawg-ipfix-on-path-telemetry-01#section-3.2.2-1>

      IPv4 header values:
        DSCP: Set to 0

      IPv6 header values:
        DSCP: Set to 0
        Hop Count: Set to 255
        Flow Label: Set to 0
        Extension Headers: None

    ¶
    <https://datatracker.ietf.org/doc/html/draft-tgraf-opsawg-ipfix-on-path-telemetry-01#section-3.2.2-2>


Looking at RFC8912 (https://www.rfc-editor.org/rfc/rfc8912.txt) we saw 
that all performance measurements set the DSCP bit to 0, then we copied 
that section.
However, thinking some more about it, we speak about perf metrics for 
traffic observed in the IPFIX metering process, typically on a router.
In IPFIX, we specify the flow record by setting up the key field and 
non-key field.
So typically, we could have flow record with DSCP value = 0, EF, etc.

As the flow record definition is flexible, we don't want to set up a new 
perf metric for potential value of key field, i.e. traffic filters ... 
whether we speak about DSCP, Flow Label, or Extension Header (from the 
example above).
Along the same lines, we don't want to create new perf metric for every 
transport protocol, if the transport protocol is specified in the flow 
record definition.
 From Table 2 on RRC8911. 
https://datatracker.ietf.org/doc/html/rfc8911#table-2, we see the 
SubTypeMethod (which should be part of the name btw) can contain ICMP, 
IP, UDP, TCP, etc.
So, from 
https://www.rfc-editor.org/rfc/rfc8911.html#name-traffic-filter, this 
section applies:

    The Traffic Filter itself depends on the needs of the metric itself
    and a balance of an operator's measurement needs and a user's need
    for privacy. Mechanics for conveying the filter criteria might be
    the BPF (Berkeley Packet Filter) or PSAMP (Packet Sampling)[RFC5475
    <https://www.rfc-editor.org/rfc/rfc8911.html#RFC5475>]Property Match
    Filtering, which reuses IPFIX[RFC7012
    <https://www.rfc-editor.org/rfc/rfc8911.html#RFC7012>]. An example
    BPF string for matching TCP/80 traffic to remote Destination net
    192.0.2.0/24 would be "dst net 192.0.2.0/24 and tcp dst port 80".
    More complex filter engines may allow for matching using Deep Packet
    Inspection (DPI) technology.

Do we correctly understand that:
     1. the traffic filter field should remain empty
        Do we want to mention something about the IPFIX flow definition?
     2. we don't include the SubTypeMethod in the name (as we have done 
in draft-tgraf-opsawg-ipfix-on-path-telemetry)


Thanks for your guidance.

Regards, Benoit