perfsonar development work

[Nicolas Simar <>]

I am suggesting to add that point to the next perfsonar meeting.
The objectives would be to define the problem, its dependencies and 
assess the different solutions.

I would foresee:
- visualisation tools: AH, ?
- LS: MG, MS
- Tools: VV, JB

Cheers,
Nicolas

Vedrin Jeliazkov wrote:
> Hi Nicolas, All,
>
> Support for perfSONAR service redundancy hasn't been much discussed up to now
> AFAIK. This might make it rather tricky for quick implementation/deployment.
>
> As far as perfsonarUI is concerned, we can imagine the following procedure:
>
> 1) LS provides info in some pre-defined and well known way whether a given
> service has several instances and where they can be reached;
>
> 2) perfsonarUI randomly selects one of those instances and sends a query; if
> the service returns some unexpected error or the query timeouts, then the
> client asks another randomly chosen service instance; this process continues
> until either the query is answered or the list of available service instances
> is exhausted;
>
> In fact, this pretty much follows the way how DNS stuff works.
>
> Kind regards,
> Vedrin
>
> Nicolas Simar 
> <>
>  wrote:
>
>
> > Hi,
> >
> > for the deployment of the MDM service on GEANT, the DANTE operation team 
> > mentioned that they would wish to have the services backup-ed on a 
> > second server.
> >
> > How can this be done? (a) for the next three months, (b) for the next 
> > release?
> >
> > 1) SSH telnet MP - two deployments each contacting the routers. No 
> > issues foreseen.
> >
> > 2) RRD MA - two deployments, how do they access the data? The original 
> > DB might be on another server, how do you synchronize those? Do you 
> > foresee another way of doing this?
> > Obviously if the server where the original data are is down, we can't do 
> > anything except if there are two instance capturing the data.
> >
> > 3) L2 status MP - two deployments, each accessing the XML file stored 
> > locally for one server, in another place for the other.
> > Obviously if the server where the original data are is down, we can't do 
> > anything except if there are two instance capturing the data.
> >
> > 4) SQL MA - two deployments. What do you suggest to do with the DB? 
> > Stored where? Any suggestions.
> >
> > 5) LS - how two instance of the LS could have the same information? (can 
> > the services register to two LSes at the same time)
> >
> > 6) How would the LS advertise the fact that there are two web-services 
> > offering the same information?
> >
> > 7) How does the visualisation tools react to have two services offering 
> > the same information? How would they react when one goes down?
> >
> > Thanks a lot in advance
> >
> > Cheers,

--
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
_____________________________________________________________________