Thanks Shahram. I think I am closer to understanding, but I have some more questions, to help address my concerns.
I have added some comments below, preceeded by PM. Thanks, Peter Shahram Davari <[email protected]> 22/03/2012 04:15 PM To "[email protected]" <[email protected]> cc "[email protected]" <[email protected]> Subject RE: [TICTOC] Updated 1588 over MPLS draf-03 Hi Peter, Thanks you for your comments. Please see my response inline. -Shahram From: [email protected] [mailto:[email protected]] Sent: Thursday, March 22, 2012 9:17 AM To: Shahram Davari Cc: [email protected] Subject: RE: [TICTOC] Updated 1588 over MPLS draf-03 Hi Shahram, There are a few points causing some confusion for me. 1) In section 19 it lists the use case that I proposed to add as an architecture diagram in section 4. " - An LSR receives MPLS encapsulated PTP messages and terminates them, while performing the OC or BC functionality" From your e-mail you seem to indicate this is not a use case. Or perhaps I have not understood your statement about "a real LSR doing BC, meaning that without terminating the LSP do the BC operation". I am only familiar with a BC that terminates PTP packets. SD> That is correct, and LSR (with respect to a PTP LSP) can’t perform OC or BC operation, since to perform OC or TC PTP LSP must be terminated and that means this is an LER and not LSR. You are correct about section 19. I have deleted that bullet that you mentioned. 2) In section 7 there was previously in -02 the below statement that peer-delay messages must be transported over single hop PTP LSPs. "SYNC and DELAY_REQ are exchanged between Master Clock and Slave Clock and MUST be transported over PTP LSPs. PDELAY_REQ and PDELAY_RESP are exchanged between adjacent PTP clocks (i.e. Master, Slave, Boundary, or Transparent) and MAY be transported over single hop PTP LSPs. If Two Step PTP clocks are present, then the FOLLOW_UP, DELAY_RESP, and PDELAY_RESP_FOLLOW_UP messages must also be transported over the PTP LSPs." Now this statement is removed in -03, as well as references to two-step PTP clocks and associated FOLLOW_UP & PDELAY_RESP_FOLLOW_UP. "SYNC and DELAY_REQ are exchanged between Master Clock and Slave Clock and MUST be transported over PTP LSPs. PDELAY_REQ and PDELAY_RESP are exchanged between adjacent PTP clocks (i.e. ordinary in Master or Slave state, boundary, or transparent clock) and MAY be transported over single hop PTP LSPs." SD> The reason the 2-Step statement was removed was that FOLLOW_UP, DELAY_RESP, and PDELAY_RESP_FOLLOW_UP don’t need to be transported over PTP LSP since they are exchanged over one hop. However the draft does not prevent one from sending them over PTP LSP. May be I can add the following Text: “If Two Step PTP clocks are present, then the FOLLOW_UP, DELAY_RESP, and PDELAY_RESP_FOLLOW_UP messages MAY be transported over the PTP LSPs” PM> There are two types of packets that are necessary to implement a p2p TC functionality, as you mention. Both of these types of packets could be event packets, and both of these types of packets are critical to enable the TC functionality. Some of these packets flow through the TC network element, for which the draft establishes a special PTP LSP. Some of these packets are generated & terminated at every TC network element (e.g. peer-delay) for which there is no recommendation in the draft, but there are a couple options. Option 1: These packets are not carried over MPLS, but over another mapping (such as proposed in draft-jobert-tictoc-ptp-link-local-00). It is not intended in this draft to specify this mechanism, even though these packets are required for the TC to operate, and are time sensitive event packets that need to be indentified easily by the ingress & egress port aswell. Since these packets are generated & terminated in the TC, but are not carried over an LSP, this portion of the TC functionality does not require the node to be considered an LER. Option 2: These packets are carried over a second PTP LSP (single hop). Would this second PTP LSP be the same type of special PTP LSP that requires the additional control aspects being proposed in this draft, in order to have a label setup to identify the event/timestamping needs of the packet? As the TC would terminate this second PTP LSP, would it not be considered to be acting as an LER? (I am not clear if generating & terminating a single hop PTP LSP qualifies the equipment as an LER, since this is not my area of expertise). For p2p deployments, the exchange & termination of packets between two LSR is still required. For the use case of p2p TC or p2p BC, each LSR must generate & terminate some of the PTP messages and process them locally. These would be the PDELAY_REQ, PDELAY_RESP and PDELAY_RESP_FOLLOW_UP. Not all PTP messages are sent through the LSR transparently in the case of p2p TC. SD> No. The idea is to follow the standard MPLS data-plane. This means we can’t terminate packets in the middle of an LSP. So in the case that you mentioned, two different PTP LSPs are needed. One for carrying end-to-end messages (between Master and Slave) and one for carrying Peer-to-peer messages. In such a case it would seem the LSR must be able to terminate a single hop PTP LSP, and this could be used for TC or BC implementation in the LSR. SD> As mentioned above BC can only be done in LER, means you have to terminate the PTP LSP and then process the 1588 message. PM> Refering back to the previous two options I have introduced, depending on which was intended to be propsed in the draft for non-flow through packets, it would seem there may be ways to address my concerns about the exclusion of BC functionality. Option 1: Using another mapping to carry PTP messages between network elements could apply to all BC messages (see draft-jobert-tictoc-ptp-link-local-00). This BC is embedded within a network element that is acting as an LSR, but the PTP itself is never carried over MPLS. This could be mentioned in the section 4 architecture, allowing the reader to more clearly understand that both BC capability and TC capability could exist within a network element acting as an LSR. At present it seems that a network element acting as an LSR could only support TC functionality, which is not the case. Option 2: The second PTP LSP (single hop) could also be used to carry all necessary BC messages. This would be, I thought, what was proposed in section 19. This could also be shown in section 4 architecture. Regards, Peter Shahram Davari <[email protected]> 21/03/2012 02:29 PM To "[email protected]" <[email protected]> cc "[email protected]" <[email protected]> Subject RE: [TICTOC] Updated 1588 over MPLS draf-03 Hi Peter, Thanks for your great comments. Please see my responses inline. Thanks Shahram From: [email protected] [mailto:[email protected]] Sent: Wednesday, March 21, 2012 10:10 AM To: Shahram Davari Cc: [email protected] Subject: Re: [TICTOC] Updated 1588 over MPLS draf-03 Hi Shahram et al, These are my comments on the draft. 1) In section 3, I would suggest this paragraph be removed. It is unclear where this problem arises as it seems the GMs are all traceable back to UTC. Many large operators have confirmed the use and distribution of only a single clock domain from core through access, in on-going discussions within ITU_T. Maybe you have some examples of asynchronous PTP clock domain deployments, and where this requirement arises? It would be interesting to know what other traceable time base may be used to distributed throughout the network, and how multiple GMs are themselves synchronized to a timebase other than UTC.. I think it would be easier to remove this paragraph than to start-up such a use case discussion. "There is also a requirement to transport PTP messages belonging to many different 1588 domains across an MPLS network,such as the case for whole sale (carrier?s carrier case)." SD> This issue was raised by others as well. Although this draft can support the mentioned functionality, but I think it is out of scope of this draft and I am going to remove this sentence. 2) In section 4, one example is provided using TCs in the LSRs. I would propse the addition of another example / use case using BCs in the LSRs. This traces back to some earlier comments (not fully vetted by MPLS routing experts) on the reflector that it seems single-hop PTP LSPs could be used between BCs when using a pure BC chain. This is one of the cases listed in section 19 "applicability statements" SD> In The BC at every hop case that you mention, the LSR is really acting as LER since it terminates hop-by-hop LSP. This is different from a real LSR doing BC, meaning that without terminating the LSP do the BC operation. So I am going to leave it as is. Please let me know whether you agree with my assessment. 3) In section 16.2, it covers on the transparent clock case. I would propose the following changes. a) "(e.g. transparent clock or boundary clock processing)" SD> Based on my answer to (2), I don’t think we need to do this change since LSR is not really doing BC. b) "After 1588 processing the packet is forwarded as a normal MPLS packet to downstream node (in the case of transparent clock processing), whereas a boundary clock terminates the 1588 packets at each node and re-generates a new packet downstream". SD> Same as above. No change. 4) In section 16.3, it covers only the transparent clock case. "(e.g. perform transparent clock or boundary clock processing)" SD> Same as above. No change 5) In section 19, there is a missing line-return between sub-bullet #2 and #3 SD> Corrected. Thanks. 6) In section 19, its unclear why the OC and BC case would not apply to non-MPLS interfaces aswell. SD> The OC and BC obviously apply to non-MPLS interface, but that is out of the scope of this document which Tries to document the 1588 over MPLS functionality. Regards, Peter Shahram Davari <[email protected]> Sent by: <[email protected]> 12/03/2012 08:17 PM To "[email protected]" <[email protected]>, "[email protected]" <[email protected]>, "'[email protected]'" <[email protected]>, CCAMP <[email protected]> cc Subject [TICTOC] Updated 1588 over MPLS draf-03 Hi, Please find attached the latest 1588 over MPLS draft (03). Since cut-off date was yesterday, we will upload this after the Paris meeting. Review is required from TICTOC, MPLS, PWE3 and CCAMP WGs, since some aspects from each of these groups are used in this draft. We will present this draft in the relevant WGs in Paris. Regards, Shahram Davari[attachment "draft-ietf-tictoc-1588overmpls-03.txt" deleted by Peter Meyer/Zarlink] _______________________________________________ TICTOC mailing list [email protected] https://www.ietf.org/mailman/listinfo/tictoc ---------- This email is confidential and may contain information that is privileged and exempt from disclosure by law. If you have received it in error, please contact the sender immediately by return email and then delete it from your system; you should not copy it or disclose its contents to anyone. Emails are not secure and cannot be guaranteed to be error free as they can be intercepted, amended, lost or destroyed, or contain viruses. Anyone who communicates with us by email is taken to accept these risks.
_______________________________________________ TICTOC mailing list [email protected] https://www.ietf.org/mailman/listinfo/tictoc
