Re: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt
Hi Stephane Many thanks for the comment. This text is a compromise to accommodate implementations of the previous version of the draft that have already been deployed. Those implementations do not send PATH-SETUP-TYPE-CAPABILITY in the OPEN. Instead, they send PCEP-SR-CAPABILITY. Hence, to interoperate with those versions, a new implementation can use the presence of PCEP-SR-CAPABILITY to infer support of SR, rather than PATH-SETUP-TYPE-CAPABILITY. In other words, this is for backwards compatibility. It is discussed explicitly in the new version of draft-ietf-pce-segment-routing. If you think the wording below can be improved for clarity, then please let me know! Cheers Jon -Original Message- From: stephane.litkow...@orange.com [mailto:stephane.litkow...@orange.com] Sent: 20 November 2017 15:32 To: Jonathan Hardwick ; pce@ietf.org Cc: MEURIC Julien IMT/OLN Subject: RE: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt Hi Jon, Thanks for the update. One comment regarding this paragraph: "If the peer has sent no PATH-SETUP-TYPE-CAPABILITY TLV, then the PCEP speaker MUST infer that the peer supports path setup using at least RSVP-TE. The PCEP speaker MAY also infer that the peer supports other path setup types, but the means of inference are outside the scope of this document." Why not enforcing here ? I mean if a PCEP speaker uses a PST that was not advertised in the capability TLV, the PST is rejected by a PCError. During the OPEN message, peers may agree on the PST to be used on the session based on the common subset. What's your opinion on that ? Brgds, Stephane -Original Message- From: Jonathan Hardwick [mailto:jonathan.hardw...@metaswitch.com] Sent: Monday, November 20, 2017 16:07 To: pce@ietf.org Cc: MEURIC Julien IMT/OLN; LITKOWSKI Stephane OBS/OINIS Subject: FW: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt Dear PCE WG This new revision of the LSP setup type draft makes the following changes. 1) Added a capability TLV for the OPEN object and rules for processing it, as discussed in the attached thread. This is to address Julien's WGLC comment that there was no way for a PCEP speaker to express that it doesn't support RSVP-TE as a path setup type. 2) Made the path setup type explicit for anything other than RSVP-TE paths (where absence of TLV implies RSVP-TE). This is to address Stephane's recent comment to the list. 3) Updated the IANA / code point text to reflect that we have had an early allocation. 4) Made some editorial fixes and clarifications. Best regards Jon -Original Message- From: Pce [mailto:pce-boun...@ietf.org] On Behalf Of internet-dra...@ietf.org Sent: 20 November 2017 14:59 To: i-d-annou...@ietf.org Cc: pce@ietf.org Subject: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt A New Internet-Draft is available from the on-line Internet-Drafts directories. This draft is a work item of the Path Computation Element WG of the IETF. Title : Conveying path setup type in PCEP messages Authors : Siva Sivabalan Jeff Tantsura Ina Minei Robert Varga Jon Hardwick Filename: draft-ietf-pce-lsp-setup-type-06.txt Pages : 10 Date: 2017-11-20 Abstract: A Path Computation Element can compute traffic engineering paths (TE paths) through a network that are subject to various constraints. Currently, TE paths are label switched paths (LSPs) which are set up using the RSVP-TE signaling protocol. However, other TE path setup methods are possible within the PCE architecture. This document proposes an extension to PCEP to allow support for different path setup methods over a given PCEP session. The IETF datatracker status page for this draft is: https://datatracker.ietf.org/doc/draft-ietf-pce-lsp-setup-type/ There are also htmlized versions available at: https://tools.ietf.org/html/draft-ietf-pce-lsp-setup-type-06 https://datatracker.ietf.org/doc/html/draft-ietf-pce-lsp-setup-type-06 A diff from the previous version is available at: https://www.ietf.org/rfcdiff?url2=draft-ietf-pce-lsp-setup-type-06 Please note that it may take a couple of minutes from the time of submission until the htmlized version and diff are available at tools.ietf.org. Internet-Drafts are also available by anonymous FTP at: ftp://ftp.ietf.org/internet-drafts/ ___ Pce mailing list Pce@ietf.org https://www.ietf.org/mailman/listinfo/pce _ Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuil
Re: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt
Sounds reasonable. Thanks -Original Message- From: Jonathan Hardwick [mailto:jonathan.hardw...@metaswitch.com] Sent: Tuesday, November 21, 2017 09:28 To: LITKOWSKI Stephane OBS/OINIS; pce@ietf.org Cc: MEURIC Julien IMT/OLN Subject: RE: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt Hi Stephane Many thanks for the comment. This text is a compromise to accommodate implementations of the previous version of the draft that have already been deployed. Those implementations do not send PATH-SETUP-TYPE-CAPABILITY in the OPEN. Instead, they send PCEP-SR-CAPABILITY. Hence, to interoperate with those versions, a new implementation can use the presence of PCEP-SR-CAPABILITY to infer support of SR, rather than PATH-SETUP-TYPE-CAPABILITY. In other words, this is for backwards compatibility. It is discussed explicitly in the new version of draft-ietf-pce-segment-routing. If you think the wording below can be improved for clarity, then please let me know! Cheers Jon -Original Message- From: stephane.litkow...@orange.com [mailto:stephane.litkow...@orange.com] Sent: 20 November 2017 15:32 To: Jonathan Hardwick ; pce@ietf.org Cc: MEURIC Julien IMT/OLN Subject: RE: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt Hi Jon, Thanks for the update. One comment regarding this paragraph: "If the peer has sent no PATH-SETUP-TYPE-CAPABILITY TLV, then the PCEP speaker MUST infer that the peer supports path setup using at least RSVP-TE. The PCEP speaker MAY also infer that the peer supports other path setup types, but the means of inference are outside the scope of this document." Why not enforcing here ? I mean if a PCEP speaker uses a PST that was not advertised in the capability TLV, the PST is rejected by a PCError. During the OPEN message, peers may agree on the PST to be used on the session based on the common subset. What's your opinion on that ? Brgds, Stephane -Original Message- From: Jonathan Hardwick [mailto:jonathan.hardw...@metaswitch.com] Sent: Monday, November 20, 2017 16:07 To: pce@ietf.org Cc: MEURIC Julien IMT/OLN; LITKOWSKI Stephane OBS/OINIS Subject: FW: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt Dear PCE WG This new revision of the LSP setup type draft makes the following changes. 1) Added a capability TLV for the OPEN object and rules for processing it, as discussed in the attached thread. This is to address Julien's WGLC comment that there was no way for a PCEP speaker to express that it doesn't support RSVP-TE as a path setup type. 2) Made the path setup type explicit for anything other than RSVP-TE paths (where absence of TLV implies RSVP-TE). This is to address Stephane's recent comment to the list. 3) Updated the IANA / code point text to reflect that we have had an early allocation. 4) Made some editorial fixes and clarifications. Best regards Jon -Original Message- From: Pce [mailto:pce-boun...@ietf.org] On Behalf Of internet-dra...@ietf.org Sent: 20 November 2017 14:59 To: i-d-annou...@ietf.org Cc: pce@ietf.org Subject: [Pce] I-D Action: draft-ietf-pce-lsp-setup-type-06.txt A New Internet-Draft is available from the on-line Internet-Drafts directories. This draft is a work item of the Path Computation Element WG of the IETF. Title : Conveying path setup type in PCEP messages Authors : Siva Sivabalan Jeff Tantsura Ina Minei Robert Varga Jon Hardwick Filename: draft-ietf-pce-lsp-setup-type-06.txt Pages : 10 Date: 2017-11-20 Abstract: A Path Computation Element can compute traffic engineering paths (TE paths) through a network that are subject to various constraints. Currently, TE paths are label switched paths (LSPs) which are set up using the RSVP-TE signaling protocol. However, other TE path setup methods are possible within the PCE architecture. This document proposes an extension to PCEP to allow support for different path setup methods over a given PCEP session. The IETF datatracker status page for this draft is: https://datatracker.ietf.org/doc/draft-ietf-pce-lsp-setup-type/ There are also htmlized versions available at: https://tools.ietf.org/html/draft-ietf-pce-lsp-setup-type-06 https://datatracker.ietf.org/doc/html/draft-ietf-pce-lsp-setup-type-06 A diff from the previous version is available at: https://www.ietf.org/rfcdiff?url2=draft-ietf-pce-lsp-setup-type-06 Please note that it may take a couple of minutes from the time of submission until the htmlized version and diff are available at tools.ietf.org. Internet-Drafts are also available by anonymous FTP at: ftp://ftp.ietf.org/internet-drafts/ ___ Pce mailing list Pce@ietf.org https://www.ietf.org/mailman/listinfo/pce ___
[Pce] Shepherd's Review of draft-ietf-pce-lsp-setup-type-06
Hi, Thank you for this new version of the I-D, it has greatly improved and clarifies former loose zones. Please find my review below. -- Abstract --- - s/traffic engineering paths (TE paths)/Traffic Engineering paths (TE paths)/ - I wonder about the expansion of "TE path" above: why not "Engineered" instead of "Engineering"? (This is global to the I-D, and beyond... RFC Editor's list includes both.) - s/label switched paths (LSPs)/Label Switched Paths (LSPs)/ -- Status --- - "https://"; was introduced in -05, but has now disappeared. -- 1. Introduction --- - s/Path Computation Element Protocol/Path Computation Element communication Protocol/ - OLD path setup type needs to be either explicitly indicated or implied in the appropriate PCEP messages (when necessary)... NEW path setup type needs to be explicitly indicated in the appropriate PCEP messages, unless RSVP-TE type is meant (omission implying this type)... -- 3. Path Setup Type Capability TLV --- - s/Initialization phase/initialization phase/ - I though the discussion on the list was about using a bitmap to identify supported PSTs: any reason why it is now a list of raw octets? - The definition of length and padding mixes the words "octet" and "bytes", depending on the field (probably to due to text coming from RFC 5440). Consistency would be welcome (the comment appears to be applicable beyond this section). -- 4. Path Setup Type TLV --- - Figure 2 explicitly includes the codepoint for the "Type" (28), the "Length" field should be treated similarly (4). - The last sentence of section 4 puzzles me. If the PATH-SETUP-TYPE TLV is not supported, the PCEP peer cares little if it was "recognized" or not. If both sub-cases are commonly handled by ignoring, an implementation always including the RSVP-TE PST will be able to interwork with an implementation knowing about the TLV without actually supporting it; the current text turns this situation into an error. (Note also that RFC 5440 does not distinguish unrecognized and unsupported in TLV processing rules.) - In case my previous comment does not fly, 3 more nits: * s/recognizes the TLV but does not support the TLV/recognizes the TLV but does not support its use/ * s/send PCErr/send a PCErr message/ * Error-Type is 2, would not 4 fit better? -- 5. Operation --- - s/Initialization phase/initialization phase/ - s/MUST infer/MUST consider/ [explicit => nothing to infer] - The text says multiple times "unless the intended PST is RSVP-TE, in which case it MAY omit the PATH-SETUP-TYPE TLV". This is inconsistent with section 4: "It is RECOMMENDED that a PCEP speaker omits the TLV if the PST is RSVP-TE." Please choose between MAY and SHOULD, and align. -- Cheers, Julien ___ Pce mailing list Pce@ietf.org https://www.ietf.org/mailman/listinfo/pce
Re: [Pce] Shepherd's Review of draft-ietf-pce-lsp-setup-type-06
Hi Julien Many thanks for the speedy review! Please see a few answers below, marked with [Jon]. (All other comments are accepted.) I will hold the document mark-ups until WGLC ends. Cheers Jon -Original Message- From: Julien Meuric [mailto:julien.meu...@orange.com] Sent: 21 November 2017 17:08 To: draft-ietf-pce-lsp-setup-t...@ietf.org Cc: pce@ietf.org Subject: Shepherd's Review of draft-ietf-pce-lsp-setup-type-06 Hi, Thank you for this new version of the I-D, it has greatly improved and clarifies former loose zones. Please find my review below. -- Abstract --- - s/traffic engineering paths (TE paths)/Traffic Engineering paths (TE paths)/ - I wonder about the expansion of "TE path" above: why not "Engineered" instead of "Engineering"? (This is global to the I-D, and beyond... RFC Editor's list includes both.) - s/label switched paths (LSPs)/Label Switched Paths (LSPs)/ -- Status --- - "https://"; was introduced in -05, but has now disappeared. -- 1. Introduction --- - s/Path Computation Element Protocol/Path Computation Element communication Protocol/ - OLD path setup type needs to be either explicitly indicated or implied in the appropriate PCEP messages (when necessary)... NEW path setup type needs to be explicitly indicated in the appropriate PCEP messages, unless RSVP-TE type is meant (omission implying this type)... -- 3. Path Setup Type Capability TLV --- - s/Initialization phase/initialization phase/ - I though the discussion on the list was about using a bitmap to identify supported PSTs: any reason why it is now a list of raw octets? [Jon] We did discuss a bit field on the list. The PST field is an 8-bit value, so a naive implementation of a bit field would use 32 bytes. This seems wasteful given we only have two values defined so far (possibly 3 with the up-coming IPv6-SR draft). I then looked at some schemes to shorten the bit field, e.g. by truncating it, but the result seemed more complicated to encode than just listing the path setup types. (I also didn't much like having a PST known both by a value and by a bit field position.) So I opted to list the PSTs. IMO it's not a problem given the low number of PSTs we expect (could be a case of famous last words!) What do you think of this tradeoff? - The definition of length and padding mixes the words "octet" and "bytes", depending on the field (probably to due to text coming from RFC 5440). Consistency would be welcome (the comment appears to be applicable beyond this section). [Jon] Let's standardize on bytes, per RFC 5440. -- 4. Path Setup Type TLV --- - Figure 2 explicitly includes the codepoint for the "Type" (28), the "Length" field should be treated similarly (4). - The last sentence of section 4 puzzles me. If the PATH-SETUP-TYPE TLV is not supported, the PCEP peer cares little if it was "recognized" or not. If both sub-cases are commonly handled by ignoring, an implementation always including the RSVP-TE PST will be able to interwork with an implementation knowing about the TLV without actually supporting it; the current text turns this situation into an error. (Note also that RFC 5440 does not distinguish unrecognized and unsupported in TLV processing rules.) [Jon] I think you are right. The same also applies to the final sentence of section 3, I believe. - In case my previous comment does not fly, 3 more nits: * s/recognizes the TLV but does not support the TLV/recognizes the TLV but does not support its use/ * s/send PCErr/send a PCErr message/ * Error-Type is 2, would not 4 fit better? -- 5. Operation --- - s/Initialization phase/initialization phase/ - s/MUST infer/MUST consider/ [explicit => nothing to infer] - The text says multiple times "unless the intended PST is RSVP-TE, in which case it MAY omit the PATH-SETUP-TYPE TLV". This is inconsistent with section 4: "It is RECOMMENDED that a PCEP speaker omits the TLV if the PST is RSVP-TE." Please choose between MAY and SHOULD, and align. [Jon] I think our intent is MAY, so we can reword the text in section 4 to "A PCEP speaker MAY omit the TLV if the PST is RSVP-TE." -- Cheers, Julien ___ Pce mailing list Pce@ietf.org https://www.ietf.org/mailman/listinfo/pce