perfsonar development work

[Roman Lapacz <>]

Jeff W. Boote wrote:
> Roman Lapacz wrote:
> > Mark Yampolskiy wrote:
> > > Hi Roman,
> > >
> > > the idea to use time seams to be the right one.
> > > However (hehe, how aren't there always some "however"s ;-)) from the 
> > > discussion in the mailing list it is not really clear whether schema 
> > > specification is already agreed and changed. Matthias and me have 
> > > agreed not to make any changes in our SW which support 
> > > undocumented/undefined/not-agreed futures of NMWG schema (ask 
> > > Loukik, he will tell you how bullheaded we are in this respect ;-))
> >
> > I checked the schema
> >
> > http://anonsvn.internet2.edu/svn/nmwg/trunk/nmwg/schema/rnc/nmbase.rnc
> >
> > and I see that using time inside parameter is fine. Jason, could you 
> > confirm?
>
> This is true, and has been true for a very, very long time. (2 years?) 
> The nmtime stuff has been there almost since the beginning of the 
> version 2 schema. nmtime.rnc was the second thing developed after 
> nmbase.rnc. (We had most of the discussions about time when dealing 
> with the version 1 schema in NMWG.)
>
> I believe the part of the schema that is not decided, is how indicate 
> transformations. The idea that some clients might not want to support 
> the full schema that an MP/MA might want to use to report - but that 
> there might be some way for the client to indicate which portions of 
> the schema it does want to support. For example - provide some way for 
> a client to indicate that they want "unix" timestamps, but not any 
> other timetype used in the response.
>
> I apologize for not getting an example down of the idea I started to 
> propose 2 weeks ago - I was co-opted for the past two weeks to write a 
> demo for our Internet2 Member meeting this week and basically had to 
> drop every thing else. I will have one ready by the Meeting next week 
> - and will send it out to the list.

It would be good to find time to discuss the schema of transformation 
stuff and container structures in parameter (see my one of previous 
emails). I asked Szymon to include these topics in the aganda.

> jeff
>
> P.S. The demo was a very good thing for perfSONAR in the Internet2 
> community. We used perfSONAR to show near-real time utilization 
> changes of a circuit while the circuit was being dynamically 
> re-provisioned for different capacities. We had FAR more people in the 
> perfSONAR session at the meeting than we ever have before.

Great. Recently I had a short presentation during the conference for 
grid community and they also were very interested in using perfSONAR.


Roman


> I'd be happy to show the demo next week (and I'll send out a link too).