perfsonar development work

[Loukik Kudarimoti <>]

Hi Eric,

I would like to clearly understand your definition of a bundle.

(a) a simple page which groups together a list of Web service products 
to be used and gives the location of where these products are available?
(b) Bunch of products packaged together using a purpose built software 
package or installer?

Currently, perfSONAR-bundle-2.0 is of type (b).

Loukik.



Eric L. Boyd wrote:
> Hi Loukik,
>
> I would prefer as a group to focus on application-community-specific 
> bundles that draw from all the services available. Different 
> communities have different needs and overlapping services have 
> different target audiences.
>
> Beyond perfSONAR 2.0, we agreed to go to the targeted bundle names, 
> not "perfSONAR 3.0". The idea was that a bundle might draw on services 
> written by multiple organizations, and the bundle is named for the 
> reason it is created. For example, the GEANT2-perfSONAR bundle might 
> be created to meet an internal GEANT2 deadline, while the 
> LHC-perfSONAR bundle would be created as a recommendation for the 
> entire LHC community. (This would differ from the MDM-appliance.) 
> Likewise, Fausto's bundle could easily be named "CACTI-perfSONAR" as 
> it is targeted at CACTI users or something else (maybe 
> CACTI-GoSC-perfSONAR), as long as it's unique and the steering 
> committee approves the name.
>
> --Eric
>
>
> Jeff W. Boote wrote:
> > Loukik Kudarimoti wrote:
> > > Joe, Fausto, all
> > >
> > > The current perfSONAR bundle contains the Command Line MP product 
> > > developed by RNP. So, certainly, it cannot be easily renamed to 
> > > geant2-perfSONAR-bundle. In fact, Guilherme Fernandes is the 
> > > developer of the perfSONAR bundle installer.
> > >
> > > I thought the perfSONAR bundle is the product tested and recommended 
> > > by the perfSONAR consortium. As perfSONAR release management, we 
> > > worked hard to make sure that the perfSONAR bundle works with 
> > > multiple programming languages and installation methods (ant, perl, 
> > > etc). Fausto's CACTIperfSONAR belongs to a new breed of products 
> > > (something that we expected). I was under the impression that Fausto 
> > > (and other developers) would like to have it included into into the 
> > > perfSONAR bundle when it was stable enough.
> > >
> > > But this doesn't have to be the case. If the current bundle were to 
> > > become geant2-perfSONAR bundle, then the current release management 
> > > will probably have to be called as geant2-perfSONAR release 
> > > management (i.e, internet2 would probably do its own release 
> > > management and maybe RNP chooses to do the same).
> >
> > Hi Loukik,
> >
> > You have a valid point. In thinking more about this, the names Joe 
> > suggested are not inclusive enough of the groups that created them. 
> > These names are more tilted toward who has done the work for the 
> > release. But, the names should probably be more about the target 
> > audience.
> >
> > The main problem I see going forward is that I am not sure how much 
> > sense it would make to put the LS or SNMP-MA that we have developed 
> > in perfSONAR-PS in the same bundle as the Java LS and Java RRD-MA.
> >
> > Somehow we need to give the users enough information to decide what 
> > is reasonable to install together. I thought the 'bundle' concept was 
> > that. Perhaps we can do something similar to what Joe said, but tie 
> > the name more to the target audience?
> >
> > These audiences could be as simple as things like:
> >
> > MDM
> > LHC
> > North American Universities
> >
> > But, this could be a bit too specific... It could end up creating far 
> > more bundles than anyone would want to deal with.
> >
> > Anyone else have a better idea?
> >
> > jeff (I'm going off-line now until Tuesday)