RE: Three GMPLS related IDs: draft-xu-ccamp-gmpls-arch-intra-dom ain-00.txt
Vishal Sharma <vishal@JasmineNetworks.com> Tue, 20 March 2001 09:11 UTC
Envelope-to: ccamp-data@psg.com
Delivery-date: Tue, 20 Mar 2001 01:14:41 -0800
Message-ID: <D72513E3A841D347A3545F0E2435E091E6E2D9@jasmine1.ad.jasminenetworks.com>
From: Vishal Sharma <vishal@JasmineNetworks.com>
To: xuyg@lucent.com
Cc: 'Zhi-Wei Lin' <zwlin@lucent.com>, mpls@UU.NET, ccamp@ops.ietf.org, ip-optical@lists.bell-labs.com
Subject: RE: Three GMPLS related IDs: draft-xu-ccamp-gmpls-arch-intra-dom ain-00.txt
Date: Tue, 20 Mar 2001 01:11:53 -0800
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Folks, I have read quite carefully through the three drafts that you referred to, and have the following comments them (they'll come one by one, in three emails). draft-xu-ccamp-gmpls-arch-intra-domain-00.txt. This is a v. good document, and provides a good overview of the various issues that need to be considered while architecting a control plane for optical networks, using MPLS/IP. Comments: i) On pp. 8, in the sentence "INE maintains LSP ID..." the LSP ID is introduced in this document for the first time. Where did this come from? Is it assumed that the document is suddenly referring to the LSP ID in RSVP-TE? If so, the connection needs to be made clear. If not, some introduction is needed before mentioning it. ii) In the same paragraph the sentences "However, it does not ... disseminate to other nodes," is unclear. What local information needn't exist on other NEs? iii) In Section 5, the phrase "Several physical attributes can be abstracted into one logical attribute," needs more elaboration, with examples. iv) On pp. 10, second last para, "The user traffic _connection service_ performance..." What is "connection service"? Some explanation needed. v) On pp. 13, under "Service Negotiation," ... "A typical example is... assignment order." What is this example saying? Seems quite unclear how it pertains to service negotiation. vi) On pp. 19, in the paragraph just above Section 5.3.2., you say "This implies that a time slot map.... routing decisions." Then, "Alternately, in the logical link ... STS-N signals." This latter phrase is not clear. What is the "logical link" representation? vii)In Section 5.4.1, the analogies to apply MPLS control plane concepts to circuit switched networks are both unclear. The first one is especially unclear, since it's not clear to the reader what exactly you seem to be drawing a parallel between? The second one, though better, is still not well said. viii) Similarly, on pp. 19, you talk again of a logical link (bundle). I think the term "logical link" needs to be explained better throughout the text. It was quite confusing. ix)In Section 5.4.1.2.2, you introduce "Connection ID" for the first time. Please define it first. x) The hierarchical label format, and its use is very _unclear_. I think if the text intends to make a case for that, the explanation just above the hierarchical label format needs to be much much better. xi) In Section 5.4.1.5, pp. 25, first paragraph. The phrase starting with "If the incoming label... NEs in parallel," is again unclear. To whom should the label be given at request time? I think what you mean is that the label should be allocated at request time, and the cross-connect (that allocated that label, to be passed to its upstream neighbor during the confirmation phase) should start configuring itself accordingly. xii) Section 5.4.1.6 on LSP tunneling and label stack, should have more explanation. How is the hierarchical label used exactly? xiii) Again in the "Delete" operation on pp. 25, you talk of LSP ID. Not clear where this comes from. RSVP-TE? xiv) Section 5.4.2, says (at last!) that "LSP ID can be defined as a source NE address + a source NE ingress label." How does a source NE have an ingress label? xv) The phrase immediately following the last one "An LSP has two sets of information ...." is again unclear. Does it mean an LSP has two sets of info. associated with it? xvi) Section 5.4.1.7 "A new LSP can be concatenated through ..." What does "concatenated through" mean?? xvii) At the end of the document, I came away with the feeling that it somehow ended abruptly. I haven't put my finger on it yet, but that is what I felt. I almost felt that more should have been there. xviii) A number of references don't seem to be referred to in the text. [LSP-HIER], [IPO-FRAM], [OSPF-OMPLS], [ISIS-OMPLS]. xix) In addition, a number of documents that probably should be referenced in this one are not referenced. For example, draft-bernstein-gmpls-optical-00.txt, makes many points germane to the subject of this draft. draft-ietf-mpls-recovery-frmwrk-02.txt, and draft-harrison-mpls-oam-00.txt, also make many points relevant to the discussions in Section 5.4.3. There exist drafts on bundling that are not mentioned here. draft-kompella-mpls-bundle-05.txt, for instance.
- RE: Three GMPLS related IDs: draft-xu-ccamp-gmpls… Vishal Sharma