perfsonar development work

[Roman Lapacz <>]

Andreas Hanemann wrote:
> Hi Roman,

Hi
> Thanks for your input. I have now added an additional interaction 
> diagram to the document that depicts the interactions. It seems to me 
> that your proposal is more flexible in the way how MAs can be combined 
> with transformation services and it also helps to keep the complexity 
> away from the client (which does not become aware that there is a 
> transformation service involved). For DFN I did not want to make the 
> MA address known to the outside world for security reasons, but I 
> think that the access can be prevented by other mecanisms, too.
>
> In summary, I would propose that your proposal is the way to go and I 
> would then change the document accordingly. Are there further opinions?

We can allow various options like:

- a client can send in a request for data also an address of TrS to 
eliminate communication between MA and LS (we may have some reasons to 
use certain instance of TrS)
- a client can send in a request for data also an address of MA where 
transformed data should be stored (transformation might be time 
consuming operation, too long for a client to wait for results; a client 
can later fetch transformed data from that MA)


Roman



> Best regards
> Andreas
>
>
> Roman Lapacz schrieb:
> > Andreas Hanemann wrote:
> > > Hi all,
> >
> > HI
> >
> > > In attachment I am providing a document about the Transformation 
> > > Service. It contains a discussion about the scope of the service and 
> > > also a description how the protocol for the service may look like. 
> > > Some comments are left in the document which result from discussions 
> > > involving few JRA1 people (Nina, Szymon and Nicolas).
> > >
> > > Comments especially for the scope and the protocol would be desired.
> >
> >
> >
> > If I remember correctly the idea of TrS was little bit different. 
> > Example scenario:
> >
> > 1. The client requests MA for data in specific format X
> > 2. MA contains data in format Y so finds (in LS) TrS which can 
> > transform from format Y into X (TrS can be also specified in the 
> > request from the client so in this case there is no need to use LS)
> > 3. MA sends the request and data to TrS
> > 4. TrS transforms data into format X
> > 5. TrS sends data back to MA (MA sends them to the client) or 
> > directly to the client
> >
> > Roman
> >
> >
> > > Best regards
> > > Andreas