perfsonar-dev - Re: L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date]
Subject: perfsonar development work
List archive
- From: Nicolas Simar <>
- To: "Matthias K. Hamm" <>, Jason Zurawski <>
- Cc: Maciej Glowiak <>, Matthias Hamm <>, Mark Yampolskiy <>, Dave Wilson <>, Martin Swany <>, Andreas Hanemann <>, Afrodite Sevasti <>, Otto Kreiter <>, Loukik Kudarimoti <>, Szymon Trocha <>, Stephan Kraft <>, Roberto Sabatino <>, Michael Enrico <>, Athanassios Liakopoulos <>, Roman Lapacz <>, GN-JRA1-list <>, "" <>
- Subject: Re: L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date]
- Date: Thu, 30 Nov 2006 14:39:57 +0000
Matthias K. Hamm wrote:
Hi Nicolas,
please find our comments inline.
Nicolas Simar schrieb:
Hi Maciej, Mark and Matthias,
Maciej Glowiak wrote:
Hi Nicolas,
Nicolas Simar napisaĆ(a):
I can help and write particular XQuery expression for you if you give me XML example of Lookup Information.
You mean that we should provide the information that will be registered in the LS.
Yes.
During yesterday conf call with JRA3-4, it was mentioned that at this stage, the discovery functionality isn't needed as the visualisation tools download all the L2 circuit information from the MA/MPs and stitch them together to display them. So it seems that at this stage, the functionality isn't needed yet for JRA3-4 application.
It was mentioned that at some point, some scalability issues may be encountered and that a discovery mechanism may be needed. However, no information explaining when the scalability issues would be encountered were given.
Can I ask Mark and Matthias to give an estimate about when their tool would need to use a discovery mechanism (because too many MA/MP, because new circuits are coming regularly and the number of MP/MAs varies greatly, or because too many circuits and a manual file doesn't do it anymore).
We don`t expect scalability issues. Following our architecture, we do not need a discovery mechanism for L2 circuit information. Howerver, we would like to have the possibility of finding the URLs to MPs and MAs by using a lookup service. This should also solve the poblem of discovering backup for unreachable MPs/MAs.
The querry you would expect to run would be the following:
"Give me the list of services where I can find L2 status information." and you would be returned a list of MAs and MPs (and domains or just the URL?), plus a list of LS to contact to get furhter information (as the LS are distributed). The you can ask recursively the question to the other LSes (filtering out the ones you have already visited).
Jason, is it how it will be expected to work?
However, as part of the perfSONAR, we need to cater for other visualisation or analysis tool (e.g. build by some NOC for punctual troubleshooting). They may have the requirement to discover:
(1) all the e2e circuits from a project, then all the MPs/MAs containing the relevant information in respect with those e2e circuits
(2) find out a particular e2e circuits based on the end-institution (and maybe the project name)
(3) find out all the information related to an end-to-end circuit (so the MAs/MPs where the information is located).
For 3, you need to know the e2e circuit name and you expect to get back, by querying the LS, the list of <domain-link, MA/M Paddress> / <inter-domain-link, MA/MP address > that are associated with the e2e circuit id. (or the <MA/MP address, list of link id>
From there, you can querry the different MA/MP and recompose the e2e path, except that at this stage, the MP can will only return you the list of all circuits when being queried.
(2) seems to require more information than just the e2e circuit id, the domain/inter-domain link id and the MA, MP where this info needs to be located. So I would set it aside for the time being.
(1) need one piece of information in addition to the triplet mentioned above: the name of the project having requested that circuit. Do we wish to add this in the information we plan to register to the LS? (who would register it, all the domain? What happens if people write the project name differently?)
Matthias and Mark, is it correct that the minimum information that a MP/MA would need to register to a LS would need to be <e2e circuit id, domain-link, MA/MP address> / <e2e circuit-id, inter-domain-link, MA/MP address> and that should be sufficient to be able to retrieve the info of a L2 circuit.
The information you have mentioned will be partly stored in cNIS. It is always a good idea to avoid having data redundancy between different services ;-)
that raises the point about the co-existence of the cNIS and the lookup service to some extend.
Although some kind of E2E link name registry is necessary (something more enhanced than collecting names on HTML pages), it is a major question if the lookup service is the right place to store this information. Escpecially, if the LS is a distributed service ;-)
What type of request would you like to do and what would you like to get as answers?
Nicolas
Cheers,
Cheers,
M&M
--
Nicolas
______________________________________________________________________
Nicolas Simar
Network Engineer
DANTE - www.dante.net
Tel - BE: +32 (0) 4 366 93 49
Tel - UK: +44 (0)1223 371 300
Mobile: +44 (0) 7740 176 883
City House, 126-130 Hills Road
Cambridge CB2 1PQ
UK
_____________________________________________________________________
- L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date], Nicolas Simar, 11/30/2006
- Re: L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date], Matthias K. Hamm, 11/30/2006
- Re: L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date], Nicolas Simar, 11/30/2006
- Re: [pS-dev] Re: L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date], Jason Zurawski, 11/30/2006
- Re: [GN2-JRA1] L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date], Martin Swany, 11/30/2006
- Re: L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date], Nicolas Simar, 11/30/2006
- Re: L2 circuit Discovery [Was: Re: JRA1-JRA3-JRA4 conf call - change of Date], Matthias K. Hamm, 11/30/2006
Archive powered by MHonArc 2.6.16.