[Pgi-wg] Sec: Discussion about agrreement of attributes
Oxana Smirnova
oxana.smirnova at hep.lu.se
Wed Apr 15 20:36:58 CDT 2009
Hi Morris,
I see nobody reacted yet on this, and actually it came with a "SPAM" label.
FWIW, some comments:
> (1) Considering interops between Grids we should provide an attribute
> that basically indicates if a user may 'belong' to a specific
> production Grid (e.g.DEISA, EGEE, NAREGI).
No, this an admittedly bad idea. A VO is a VO, a Grid is a Grid. Several
Grids may serve the same VO, it already happens. Even if it was not
ambiguous, I do not even see where could you place this attribute, and
what possibly can make use of it?
> Currently, VOs seem to be the top hierarchy - however I think it
> would be good to bring up attributes for that to distinguish at
> policy level which elements of policies should be used, e.g. users of
> TeraGrid in DEISA have a special policy, etc.
This is something to be implemented in the respective Policy Decision
Point services of these Grids, or whatever analogous service they
utilize. It has nothing to do with a VO. A VO may pre-date any Grid, you
can't expect the VO attributes being all re-shuffled every time a new
Grid appears or an old one gets renamed.
> (2) DEISA view, addressing the question of Etienne slightly), taking
> my talk of OGF25 into account if you would like to know more:
>
> Talk about how users can involved with DEISA:
> http://www.ogf.org/OGF25/materials/1557/2009-03-02_DEISA_DECI_Riedel.pdf
>
I don't see what is the question here? Fine, DEISA has this
differentiation between two different kinds of VOs, so what is the
implications for others, or for standards?
> (3) Agreements between attribute differences between job management
> and data management
What attributes should be different, and why? Do you think a user can
have different privileges regarding job and data management, for
example? How will you ensure consistency?
> (4) What do others think about the FQANs and its use - some critics
> or missing links I maybe have not talked about yet?
*I* think it is extremelly VOMS-specific term, and as such is internal
to VOMS. I wouldn't spend time on profiling FQANs.
Cheers,
Oxana
More information about the Pgi-wg
mailing list