Re: [OSPF] New Version Notification for draft-acee-ospf-rfc6506bis-01.txt

[Acee Lindem <[e-mail]>]

Hi Marek,

On Jun 12, 2013, at 5:23 PM, Marek Karasek (mkarasek) wrote:

> Hi Acee,
> 
> Please see %MK in your lines:
> 
> -----Original Message-----
> From: Acee Lindem [mailto:[e-mail]] 
> Sent: Wednesday, June 12, 2013 10:59 PM
> To: Marek Karasek (mkarasek)
> Cc: [e-mail]
> Subject: Re: [OSPF] New Version Notification for draft-acee-ospf-rfc6506bis-01.txt
> 
> Hi Marek,
> I thought about this and I think we need to modify the behavior in transition mode for this to provide a smooth transition. An OSPFv3 router in transition mode needs to "be conservative in what it sends and liberal in what it accepts". Hence, in transition mode, the router would accept packet with and without an authentication trailer but would not include the authentication trailer in this mode. If you don't do it this way, you'll have a deployment window problem since routers in transition mode will still be sending packets that won't be accepted by routers that have not yet been converted - they'll be dropped due to checksum errors. 
> 
> %MK they should not drop packets, checksums are populated, see:
>>> - For OSPFv3 packets to be transmitted in deployment mode, the OSPFv3 
>>> header checksum and LLS data block checksum is computed and written in 
>>> the packets.

Ok - I thought of this alternative but missed that you included it. At first, I thought the above would be simpler since it only modified the receiving side. 



> 
> If you don't include the authentication header, you just need to:
> 
>       1. Convert all the nodes in the routing domain to OSPFv3 authentication transition mode. There will be no timing dependencies.
>       2. Convert routers to send at the authentication trailer in any order. 
> 
> %MK what happens if router with AT receives packet without AT (from router doing transition)? I think it will be dropped. That's the issue.

Yes - I can see how taking doing both authentication and checksums could be simpler on multi-access links. We can stay with this. 

Thanks,
Acee 


> 
> My idea (already tested) was:
> 1) start deployment in "transition" mode.
> Transitioning routers will add AT.
> 1a) Routers without AT will ignore AT and verify CRC as usual. Note CRC is computed so packet is OK for them. 
> 1b) Routers running AT transition or AT will verify AT. Router running AT transition will not drop if problem with AT, they will verify also CRC and if OK packet is accepted.
> 2) Say all or most of routers are in "transition" mode. There will be show command option to verify that AT works OK since AT is verified, but packet is not dropped if AT NOK or not present. AT incompatible neighbors may be pointed out and fixed.
> 3) If all looks OK, you can move routers to standard AT mode at any time. They will stop computed CRC, it's not needed anymore.
> 
> Thanks marek
> 
> 
> Agree? 
> 
> Thanks,
> Acee 
> 
> 
> On 6/11/13 6:43 AM, "Acee Lindem" <[e-mail]> wrote:
> 
>> Hi Marek,
>> I've thought about it and this would be compatible with the rest of the 
>> draft. It would be useful if incremental deployment is a concern. I 
>> have no objection to adding this. Any other opinions?
>> 
>> Thanks,
>> Acee
>> On Jun 11, 2013, at 9:26 AM, Marek Karasek (mkarasek) wrote:
>> 
>>> Hi Acee,
>>> 
>>> I support bis version as well.
>>> 
>>> I have one more suggestion though for this paragraph:
>>> 
>>>  In support of uninterrupted deployment, an OSPFv3 router implementing
>>>  this specification MAY implement a transition mode where it includes
>>>  the Authentication Trailer in transmitted packets but does not verify
>>>  this information in received packets.  This is provided as a
>>>  transition aid for networks in the process of migrating to the
>>>  authentication mechanism described in this specification.
>>> 
>>> 
>>> Can it be explicitly added how to work with checksums in the 
>>> transition (or deployment) mode? I suggest adding:
>>> 
>>> - For OSPFv3 packets to be transmitted in deployment mode, the OSPFv3 
>>> header checksum and LLS data block checksum is computed and written in 
>>> the packets.
>>> - For packets received in deployment mode which include an OSPFv3 
>>> Authentication Trailer, OSPFv3 header checksum verification MUST be 
>>> omitted.
>>> - For packets received in deployment mode which do not include an
>>> OSPFv3 Authentication Trailer, OSPFv3 header checksum and LLS data 
>>> block checksum are verified.
>>> 
>>> 
>>> Thanks marek
>>> 
>>> 
>>> -----Original Message-----
>>> From: [e-mail] [mailto:[e-mail]] On Behalf 
>>> Of Acee Lindem
>>> Sent: Tuesday, June 11, 2013 1:35 PM
>>> To: Michael Barnes (mjbarnes); [e-mail]
>>> Subject: Re: [OSPF] New Version Notification for 
>>> draft-acee-ospf-rfc6506bis-01.txt
>>> 
>>> Thank Michael - Does anyone else support this work? I think it will 
>>> help ensure compatibility between implementations. I would have 
>>> expected at least those who submitted the corrected errata to support the draft.
>>> Thanks,
>>> Acee
>>> 
>>> On 6/6/13 1:12 PM, "Michael Barnes" <[e-mail]> wrote:
>>> 
>>>> I agree these are good changes. Acee, please move forward with this 
>>>> draft.
>>>> 
>>>> Thanks,
>>>> Michael
>>>> 
>>>> On 05/09/2013 11:03 AM, Acee Lindem wrote:
>>>>> There have been a couple errata filed on RFC 6505 (authors copied).
>>>>> As a service to the  OSPF community and in an effort to ensure 
>>>>> interoperable OSPFv3 authentication  trailer implementations, I 
>>>>> have produced a BIS draft. The changes are listed in  section 1.2:
>>>>> 
>>>>> 1.2.  Summary of Changes from RFC 6506
>>>>> 
>>>>>   This document includes the following changes from RFC 6506
>>>>> [RFC6506]:
>>>>> 
>>>>>   1.  Sections 2.2 and 4.2 explicitly state the Link-Local Signalling
>>>>>       (LLS) block checksum calculation is omitted when an OSPFv3
>>>>>       authentication is used.  The LLS block is included in the
>>>>>       authentication digest calculation and computation of a checksum
>>>>>       is unneccessary.  Clarification of this issue was raised in an
>>>>>       errata.
>>>>> 
>>>>>   2.  Section 4.5 includes a correction to the key preparation to use
>>>>>       the protocol specific key (Ks) rather than the key (K) as the
>>>>>       initial key (Ko).  This problem was also raised in an errata.
>>>>> 
>>>>>   3.  Section 4.5 also includes a discussion of the choice of key
>>>>>       length to be the hash length (L) rather than the block size 
>>>>> (B).
>>>>>       The discussion of this choice was included to clarify an issue
>>>>>       raised in a rejected errata.
>>>>> 
>>>>>   4.  Section 4.1 indicates that sequence number checking is 
>>>>> dependent
>>>>>       on OSPFv3 packet type in order to account for packet
>>>>>       prioritization as specified in [RFC4222].  This was an omission
>>>>>       from RFC 6506.
>>>>> 
>>>>> 
>>>>> I would like to quickly move this to an OSPF WG document and begin 
>>>>> the review process. I'm now soliciting feedback on OSPF WG adoption.
>>>>> 
>>>>> Thanks,
>>>>> Acee
>>>>> 
>>>>> 
>>>>> On May 9, 2013, at 1:43 PM, <[e-mail]>
>>>>> wrote:
>>>>> 
>>>>>> 
>>>>>> A new version of I-D, draft-acee-ospf-rfc6506bis-01.txt has been 
>>>>>> successfully submitted by Manav Bhatia and posted to the IETF 
>>>>>> repository.
>>>>>> 
>>>>>> Filename:	 draft-acee-ospf-rfc6506bis
>>>>>> Revision:	 01
>>>>>> Title:		 Supporting Authentication Trailer for OSPFv3
>>>>>> Creation date:	 2013-05-09
>>>>>> Group:		 Individual Submission
>>>>>> Number of pages: 25
>>>>>> URL:         
>>>>>> http://www.ietf.org/internet-drafts/draft-acee-ospf-rfc6506bis-01.txt
>>>>>> Status:      
>>>>>> http://datatracker.ietf.org/doc/draft-acee-ospf-rfc6506bis
>>>>>> Htmlized:    
>>>>>> http://tools.ietf.org/html/draft-acee-ospf-rfc6506bis-01
>>>>>> Diff:        
>>>>>> http://www.ietf.org/rfcdiff?url2=draft-acee-ospf-rfc6506bis-01
>>>>>> 
>>>>>> Abstract:
>>>>>>  Currently, OSPF for IPv6 (OSPFv3) uses IPsec as the only mechanism
>>>>>>  for authenticating protocol packets.  This behavior is different  
>>>>>> from
>>>>>>  authentication mechanisms present in other routing protocols  
>>>>>> (OSPFv2,
>>>>>>  Intermediate System to Intermediate System (IS-IS), RIP, and 
>>>>>> Routing
>>>>>>  Information Protocol Next Generation (RIPng)).  In some  
>>>>>> environments,
>>>>>>  it has been found that IPsec is difficult to configure and maintain
>>>>>>  and thus cannot be used.  This document defines an alternative
>>>>>>  mechanism to authenticate OSPFv3 protocol packets so that OSPFv3  
>>>>>> does
>>>>>>  not only depend upon IPsec for authentication.  This document
>>>>>>  obsoletes RFC 6506.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> 
>>>>>> The IETF Secretariat
>>>>>> 
>>>>> 
>>>>> _______________________________________________
>>>>> OSPF mailing list
>>>>> [e-mail]
>>>>> https://www.ietf.org/mailman/listinfo/ospf
>>>>> 
>>>> _______________________________________________
>>>> OSPF mailing list
>>>> [e-mail]
>>>> https://www.ietf.org/mailman/listinfo/ospf
>>> 
>>> _______________________________________________
>>> OSPF mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/ospf
>> 
>> _______________________________________________
>> OSPF mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/ospf
>