Re: [Pce] Re: Progress on draft-ietf-pce-discovery-reqs-01.txt

Hi Stephen,

On Sep 8, 2005, at 5:10 PM, Stephen Morris wrote:

> Hi JP
> I'd be happy to help with the MIB work.

good, thanks.

JP.

> Just one reply
> below :-).
> Best wishes
> Stephen
>
> --- JP Vasseur <jvasseur@cisco.com> wrote:
>
>
>> Hi,
>>
>> Adding a few comments.
>>
>> On Sep 7, 2005, at 6:00 AM, Adrian Farrel wrote:
>>
>>
>>> Hi,
>>>
>>>
>>>
>>>> A network manager would probably want NMS data on
>>>>
>> all
>>
>>>> of the PCEs in the network. It's a useful
>>>>
>> mechanism in
>>
>>>> discovery to let the managed entities (PCEs in
>>>>
>> this
>>
>>>> case) notify the NMS about changes (such as
>>>> capability).
>>>>
>>
>> As pointed out in other emails related to this
>> thread, there is
>> nothing preventing the NMS to listen the PCE
>> discovery protocol.
>>
>> That said,
>> (1) NMS will undoubtedly require more data that what
>> will/should be
>> provided by PCE discovery (since PCCs will not need
>> as much data as NMS)
>> (2) Some NMS may not support the PCE discovery
>> protocol.
>>
>> We're seeking for volunteers to work on future MIBs
>> ... Could we
>> count you in ?
>>
>>
>>>> This is particularly so in cases where
>>>> there are many such entities.
>>>>
>>>>
>>>
>>> Yes, but let's not get things out of proportion.
>>>
>> There will always
>>
>>> be at
>>> least as many network switches/routers as there
>>>
>> are PCEs in any
>>
>>> network
>>> (such is the nature of the architecture).
>>>
>>> How do switches/routers currently keep their
>>>
>> capabilities (i.e.
>>
>>> inventory)
>>> up-to-date with NMS? Why shouldn't PCEs use the
>>>
>> same technique?
>>
>>>
>>>
>>>
>>>> So, I think we have 2 related points here:
>>>> (1) External entities (e.g., NMS) should be able
>>>>
>> to
>>
>>>> listen in on PCE changes, including detecting new
>>>>
>> PCEs
>>
>>>> as they appear. This reduces polling.
>>>>
>>>>
>>>
>>> Agreed. If the NMS is capable of listening, there
>>>
>> is nothing in any
>>
>>> of the
>>> suggestions that prohibits it. However, we should
>>>
>> not, I think,
>>
>>> make it
>>> compulsory for an NMS to listen to the PCE
>>>
>> discovery protocols
>>
>>> since this
>>> may be operationally impossible in some
>>>
>> configurations.
>>
>
> Good point. Often, the NMS will have its own of
> discovery software that relies on a heady cocktail of
> MIBs, ping, etc. I guess my main point is that the PCE
> discovery requirements should try to make life a
> little easier on the NMS :-).
>
>
>>>
>>>
>>
>> See my comment (2) above.
>>
>>
>>>
>>>
>>>> (2) PCEs should be able to assist the NMS in
>>>> discovering them and their capabilities
>>>>
>>>>
>>>
>>> Certainly if we allow capabilities exchange as
>>>
>> part of the PCC-PCE
>>
>>> protocol, this would be possible. The NMS would
>>>
>> simply behave as a PCC
>>
>>> (which it will probably want to do anyway for
>>>
>> computational purposes).
>>
>>>
>>>
>>
>> I'm not really convinced here ... If the NMS does
>> not require path
>> computation (which is likely to be the case in
>> various
>> circumstances), having the NMS acting as a PCC (even
>> if the PCE
>> communication protocol support detailed capability
>> discovery), I'd
>> prefer the NMS to rely on MIBs.
>>
>> Thanks.
>>
>> JP.
>>
>>
>>>
>>>
>>>> I'm basically trying to reduce the amount of
>>>> discovery-related traffic. The latter often gives
>>>> network management a bad name :-).
>>>>
>>>> Best wishes
>>>> Stephen
>>>>
>>>> --- Adrian Farrel <adrian@olddog.co.uk> wrote:
>>>>
>>>>
>>>>
>>>>> Hi,
>>>>>
>>>>> Jean-Louis is right; the requirement is focused
>>>>>
>> on
>>
>>>>> allowing PCEs to convey
>>>>> information to PCCs.
>>>>>
>>>>> Another entity would be able to "listen in" and
>>>>> gather information for its
>>>>> own use, but such a requirement is out of scope
>>>>>
>> for
>>
>>>>> this document.
>>>>>
>>>>> I guess it depends on the purpose of the
>>>>>
>> function
>>
>>>>> you are suggesting. For
>>>>> example, if you want the NMS to be able to track
>>>>>
>> the
>>
>>>>> status of the PCE,
>>>>> this would probably be a perfectly adequate
>>>>> technique. However, it does
>>>>> rely on the NMS being able to listen in. That
>>>>>
>> is:
>>
>>>>> suppose the solution to
>>>>> PCE discovery is an opaque LSA in OSPF; you
>>>>>
>> would
>>
>>>>> need the NMS to be able
>>>>> to participate (at least passively) in the OSPF
>>>>> exchanges. Since the NMS
>>>>> may be remote through a distinct management
>>>>>
>> plane
>>
>>>>> network, this might not
>>>>> be possible.
>>>>>
>>>>> The alternative solution to gathering status
>>>>> information at the NMS would
>>>>> appear to be unchanged from the way network
>>>>>
>> devices
>>
>>>>> report status
>>>>> information today. From an IETF standardisation
>>>>> point of view, this
>>>>> requires MIB modules to allow the NMS to read
>>>>> information and to let the
>>>>> PCE report asynchronous Notifications.
>>>>>
>> Obviously,
>>
>>>>> piles of other
>>>>> proprietary or standard mechanisms exist outside
>>>>>
>> the
>>
>>>>> IETF.
>>>>>
>>>>> Cheers,
>>>>> Adrian
>>>>>
>>>>> ----- Original Message -----
>>>>> From: "LE ROUX Jean-Louis RD-CORE-LAN"
>>>>> <jeanlouis.leroux@francetelecom.com>
>>>>> To: "Stephen Morris" <stephenbjm@yahoo.com>;
>>>>> <dpapadimitriou@psg.com>;
>>>>> <dimitri.papadimitriou@alcatel.be>; "JP Vasseur"
>>>>> <jvasseur@cisco.com>
>>>>> Cc: "Ting_Wo Chung" <ting_wo.chung@bell.ca>;
>>>>> <pce@ietf.org>; "Richard
>>>>> Rabbat" <richard.rabbat@us.fujitsu.com>;
>>>>> "raymond.zhang"
>>>>> <raymond.zhang@bt.infonet.com>; "Paul Mabey"
>>>>> <paul.mabey@qwest.com>
>>>>> Sent: Monday, September 05, 2005 6:49 AM
>>>>> Subject: RE: [Pce] Re: Progress on
>>>>> draft-ietf-pce-discovery-reqs-01.txt
>>>>>
>>>>>
>>>>> Hi Stephen,
>>>>>
>>>>> If the "interested entity" is a PCC then the
>>>>>
>> answer
>>
>>>>> is YES.
>>>>>
>>>>>
>>>>> Regards
>>>>>
>>>>> JL
>>>>>
>>>>>
>>>>>
>>>>>> -----Message d'origine-----
>>>>>> De : pce-bounces@lists.ietf.org
>>>>>> [mailto:pce-bounces@lists.ietf.org] De la part
>>>>>>
>> de
>>
>>
> === message truncated ===
>
>
>
>
> ___________________________________________________________
> How much free photo storage do you get? Store your holiday
> snaps for FREE with Yahoo! Photos http://uk.photos.yahoo.com
>

_______________________________________________
Pce mailing list
Pce@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/pce