perfsonar development work

["Jeff W. Boote" <>]

On May 14, 2008, at 11:30 AM, Joe Metzger wrote:

> Maxim,
> I am not sure I agree with the statement that "metadata is supposed  
> to be very static".
> In the dynamic virtual circuit environment, metadata for circuit  
> utilization is dynamic.

In this case - the information about the circuit is now data and we  
should build the abstraction model to support that information in that  
way. The reason to have the metadata/data separation is due to the  
time relevance of the information, so the implementation can deal with  
time-invariant content in a different way from time-variant content.  
(Otherwise we would just call all of it data. ;) )

jeff

> But I do think starting to think about this is a good idea!
>
> --Joe
>
>
> On May 14, 2008, at 10:25 AM, Maxim Grigoriev wrote:
>
> > Hello,
> > There is  a need to provide information about the state of metadata  
> > stored
> > at MA. The metadata is supposed to be very static but some services  
> > has very
> > substantial md contents.  There is only one informational call -  
> > Echo which
> > can be utilized in more wider sense. It gives you information about
> > availability of the service but if it would give you the timestamp  
> > of the
> > last time when metadata were updated/changed then any client would  
> > be much
> > more effective in  sending MDK request once, caching it and reusing  
> > already
> > cached therefore minimizing message exchange.
> > Cheers,
> >          Maxim