Re: [Detnet] Magnus Westerlund's Discuss on draft-ietf-detnet-mpls-11: (with DISCUSS)

[Janos Farkas <[e-mail]>]

Hi Magnus,

I think the parameter you are looking for is there in section  4.2.2.3. Packet Ordering Function Processing (https://tools.ietf.org/id/draft-ietf-detnet-mpls-11.html#pof-requirements) just encoded a different way; in particular in the number of packets:
“Note that an implementation MAY wish to constrain the maximum number of out of order packets that can be processed, on platform-wide or per flow basis. Some implementations MAY support the provisioning of this number on either a platform-wide or per flow basis. The number of out of order packets that can be processed also impacts the latency of a flow.”

Ultimately, the latency requirements of flows have to be met. The maximum packet size of a DetNet flow is known, see item c) MaxPayloadSize in https://tools.ietf.org/html/draft-ietf-detnet-flow-information-model-10#section-5.5 and max-payload-size in https://tools.ietf.org/html/draft-ietf-detnet-yang-07. 
So, maximum POF buffer size to be applied for a flow can be determined; can be converted to time and taken into account during latency engineering of a DetNet network; thus bounded latency can be provided.

I fully agree with you that this needs to be taken into account. Together with what the set of documents that specify DetNet provide, I think it is fine what we have in the MPLS data plane draft. When it comes to configuration - as you pointed out - I think the job we need to do is to pay attention to do it right in the YANG draft, which is still under development.

Regards,
Janos

-----Original Message-----
From: Magnus Westerlund <[e-mail]> 
Sent: Wednesday, September 9, 2020 4:26 PM
To: [e-mail]; magnus.westerlund=[e-mail]
Cc: [e-mail]; [e-mail]; [e-mail]; [e-mail]; [e-mail]
Subject: Re: Magnus Westerlund's Discuss on draft-ietf-detnet-mpls-11: (with DISCUSS)

Hi,

Have thought about the discuss a bit more and think I have come to kernel of the issue related to bounded latency in this context.

So the POF must be able to buffer for a certain period to be able to reorder.
The Detnet very prominently says it can provide bounded latency and I think the only shortcoming is the consideration of the POF and how take that into account.

So https://datatracker.ietf.org/doc/draft-ietf-detnet-data-plane-framework/:

3.5.2.2.  Path Differential Delay

   In the preceding example, proper working of duplicate elimination and
   reordering of packets are dependent on the number of out-of-order
   packets that can be buffered and the delay difference of arriving
   packets.  DetNet uses flow-specific requirements (e.g., maximum
   number of out-of-order packets, maximum latency of the flow) for
   configuration of POF-related buffers.  If the differential delay
   between paths is excessively large or there is excessive mis-ordering
   of the packets, then packets may be dropped instead of being
   reordered.  Likewise, PEF uses the sequence number to identify
   duplicate packets, and large differential delays combined with high
   numbers of packets may exceed the ability of the PEF to work
   properly.

I think this makes it clear that there are a requirement on the POF also for MPLS to handle these aspects. So the main question from my perspective for the WG to consider I think is if there is a solution where the POF can be configured with a maximum latency incurred by the POF or if there is need for additional meta data (timestamp) in the MPLS encapsulation to be able to provide this function. 

Assuming that links and forwarding and other packet buffers can be bounded then the reordering should be possible to be simply configured with a maximum latency in reordering so that packet that arrives are released within that time frame to achieve bound latency. However, I am not certain that I have considered all aspects of inter packet arrival and its impact also on the actual reordering window. 

Thus, resolving this aspect of the discuss may be as simple as mandating a configuration of the maximum latency incurred by the POF so that the latency of the MPLS network in dealing with a particular S-Label can be bounded. 

Cheers

Magnus Westerlund 


----------------------------------------------------------------------
Networks, Ericsson Research
----------------------------------------------------------------------
Ericsson AB                 | Phone  [phone]
Torshamnsgatan 23           | Mobile [phone]
SE-164 80 Stockholm, Sweden | mailto: [e-mail]
----------------------------------------------------------------------