Re: [Pce] Update on draft-ietf-pce-policy-enabled-path-comp

["Adrian Farrel" <[e-mail]>]

Christian,

Many thanks for a good review and useful comments.

Can I assume that there are no further comments coming specifically from the 
IPSphere Forum?

Authors, can you tell me whether the scope of comments here are large enough 
to justify a new revision before working group last call? It seems to me 
that the answer should probably be "yes", but I'm open to persuasion.

Thanks,
Adrian

----- Original Message ----- 
From: "JACQUENET Christian RD-DDEV-REN" 
<[e-mail]>
To: "JP Vasseur" <[e-mail]>; <[e-mail]>; "Lou Berger" 
<[e-mail]>; "Igor Bryskin" <[e-mail]>; "Dimitri 
Papadimitriou" <[e-mail]>; <[e-mail]>
Cc: "Thomas Walsh" <[e-mail]>
Sent: Tuesday, October 09, 2007 12:47 PM
Subject: RE: [Pce] Update on draft-ietf-pce-policy-enabled-path-comp


Dear all,

Please find herein my comments, which are *not* on behalf of the IPSF - only 
personal :-)
1. General:

I think this is a very useful document that clearly identifies the 
advantages of policy-based path computation management in PCE environments. 
Nevertheless, I would encourage a refined organization of the document, 
which would distinguish between the framework (current sections 2.1, 4 and 5 
basically), the requirements (section 3, but also part of the text that 
relates to scenarios) and use cases or scenarios (sections 2.2.x, a bit of 
sections 7 and 8).

2. Section 2.1, page 5:

The introductory sentence of the third paragraph implicly indicates that PCE 
may very well behave as either a PDP (makes decisions to be applied by PCC 
clients for example) or a PEP (applies decisions possibly made by PCC 
clients and/or other PCE components). I think this is one of the very key 
notions introduced by this draft, and would therefore suggest an 
introductory section that would elaborate on this heuristic, possibly after 
the reminder about basic concepts of policy-based management.

3. Section 2.1, page 6:

The last sentence of the first paragraph is slightly confusing: I don't 
think the "or" is exclusive, that is, provider-defined policies might very 
well be real time.

4. Section 2.2.3, page 11:

What's a policy-enabled PCC? A PCC that either embeds a PEP, a PDP or both? 
I think this should be elaborated, possibly in the Terminology section. Note 
also that the last sentence of the page mixes the notions of "making 
decision" and "apply user or service-specific policies": since the text 
introduces a kind of causality effect (the PEP embedded in the PCC enforces 
user or service-specific policies, which seems to result in soliciting the 
PDP embedded in the PEP to make decisions about the scope of constraints 
that need to be taken into account for path computation purposes), I would 
suggest a kind of flow diagram that could possibly elaborate on the 
corresponding chronology.

5. Section 2.2.3, page 12:

Related to the comment above, the sentence "when deciding..." of the first 
paragraph may also suggest the activation of a LPDP capability. I would also 
elaborate on the possible interactions between the PEP, the PDP and the 
possible LPDP capabilities that may be embedded in the PCC, assuming a 
differentiation "a la COPS Client-Type". Who's the destinee of the path 
computation request in the "once the constraints..." sentence of the same 
paragraph - the PCE?

In the third paragraph, it seems that (1) The PCC embeds a PDP that makes 
decisions about constraint-defined policies, (2) Sends a path computation 
request that conveys the aforementioned policy information towards the PCE, 
whose embedded PEP (3) Applies the corresponding decisions: is this 
statement correct? If so, I don't understand why the PCE may "decide to 
reject the request": that is, from a policy management standpoint, only the 
PEP of the PCE is solicited in this context, which means that the PEP may 
not be capable of enforcing the decision, but I don't think the PEP is 
entitled to *reject* the decision, as part of a decision making process. 
Maybe it's a wording issue.

What would be the conditions to send path computation requests to more than 
one PCE, aside from an inter-domain context? I'm not sure I understand why 
the response sent by the PCE-PEP back to the requesting PCC-PDP may indicate 
*policies*. Unless the document refers to a kind of PCE-inferred feedback 
PIB (RFC 3571)? In that case, I would elaborate.

6. Section 3, page 15:

The notion of trusted nodes (as introdcued in the third paragraph) should 
deserve some elaboration, IMHO. The security considerations of this page 
should either be extended (e.g. preservation of the confidentiality of the 
information that will be exchanged between a PDP and a set of PDPs), or 
reference the "true" Security Considerations section of the document, which 
BTW, rightfully elaborates on such issues.

7. Section 5, page 18:

I don't understand why a protocol like COPS couldn't be used in the case 
where PEP and PDP are colocated. In any case, a protocol is required to 
convey the exchanges between the PEP and the PDP...

In addition (last line of the page), it's very unlikely that SNMP will be 
used to retrieve information from a PIB.

8. Section 5, page 19:

I would suggest some elaboration on the kind of information any given PEP 
(PCC or PCE) may receive from another *PEP* - is this a typo (i.e. read PDP 
instead), or does this relate to information different from "policy-related" 
information or else? In a COPS context, multiple Client-Types can certainly 
reside and coexist within a PEP, but I'm not sure I understand why PEPs 
would exchange information.

Also, the following paragraph ("any given policy...") is a bit confusing to 
me: I don't understand this notion of partial interpretation, and would 
rather suggest another wording - enforcement instead of interpretation. 
Besides, I'm not sure this paragraph helps in understanding the roles played 
by PEPs and PDPs.

9. Section 6.1., page 20:

Why *all* PC policy types used in the net *must* be applied? It's a 
"Client-Type" specific thing, IMHO.

10. Section 6.1., page 21:

In the first paragraph, I don't understand why the PCE selection should be 
static. Even if the policy enforcement is restricted to the PCE (for path 
computation purposes), nothing prevents a PEP embedded in a PCC to support 
the relevant "Client-Type", hence yielding the dynamic identification of the 
corresponding PCE, by means of the COPS-like OPEN message at PCC bootstrap, 
for example.

In the third paragraph, the "must" of the first sentence is too strong, 
IMHO, unless the sentence means that any kind of policy must "encouarge" the 
support of the corresponding "Client-Types" in the PCC-PEP. If so, I would 
rephrase.

11. Section 6.2., page 22:

I don't understand why matching repositories would result in a single PC 
repository, unless further elaboration on such match is provided.

12. Section 6.3, page 23:

Does the figure relate to PCE instead of PCC (same for figure 12)?

13. Section 7.1., page 25:

What's an "objection function"?

14. Typos:

TED = Traffic Engineering Database?

SRLG = Shared Risk Link Group?

Page 16, provide the normative reference associated to LDAPv3.

Page 17, remove the "s" of the very last word of the page.

Cheers,

Christian.


________________________________

De : JP Vasseur [mailto:[e-mail]]
Envoyé : lundi 8 octobre 2007 22:36
À : [e-mail]; Lou Berger; Igor Bryskin; Dimitri Papadimitriou; 
[e-mail]
Cc : Thomas Walsh; JACQUENET Christian RD-DDEV-REN
Objet : Fwd: [Pce] Update on draft-ietf-pce-policy-enabled-path-comp


Dear WG,

Christian Jacquenet on behalf of IPSphere proposed to make several comments 
on this document.
Christian, could you please send them to the list ?

As agreed, we're still planning to issue a WG LC after this round of 
discussion.

Thanks.

JP.


Begin forwarded message:


From: JP Vasseur <[e-mail]>
Date: September 18, 2007 2:10:12 PM EDT
To: [e-mail], Lou Berger <[e-mail]>, Igor Bryskin 
<[e-mail]>, Dimitri Papadimitriou <[e-mail]>, 
[e-mail]
Cc: Thomas Walsh <[e-mail]>, JACQUENET Christian RD-TCH-REN 
<[e-mail]>
Subject: [Pce] Update on draft-ietf-pce-policy-enabled-path-comp

Dear WG,

>From the WG minutes of the IETF-69 meeting:


12) Update on Policy-Enabled Path Computation Framework
draft-ietf-pce-policy-enabled-path-comp-01.txt (Lou - 5mn) [110]


Lou> ready for Last Call
JP> It sounds ready. Suggestion: offer IPSphere to have a look at the doc 
before last call. JP will be the point
of contact. Asked Ross whether this is a good idea, and Ross agreed. Once 
comments are received, last call.
Adrian> OK, but this is not an indefinite consultation. We want to be able 
to move ahead and complete the I-D
relatively soon.

We have contacted IPSphere and RA WG of IPSphere should provide us some 
feed-back by
mid-October. I have copied Tom Walsh and Christian Jacquenet (chair of the 
RAWG) IPSphere.

Thanks.

JP.




_______________________________________________
Pce mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/pce