The consolidated view on the common (or not) registry:
-) we do see a need for a common registry of high level resources, which
is manually maintained: to set up monitoring in a reliable way.
 Currently the working solution is GOCDB.

-) we don't see any important requirement for a common to all middlewares
top-level information cache. The amount of free space/available space on a system or whether a version of the software
is actual is checked by Nagios monitoring probes so we have an
information on status and also alarms. Providing information for some
statistical purposes can be done using scripts/GRIS2.0/GRIS2.0 + plugin
in a much easier way then by developing (and maintaining) a new
component.

A common API sounds better for us than a new service common to all middlewares.

Whatever the decision is for brokering purposes the common information system idea should prove it
will not limit the functionality available to users to an "intersection"
of features available in all middlewares. This would make users never
turn to the new solution.

-) if somebody (e.g. EMI) is going to provide us with a top-level
information cache, we can happily use it. However if it is going to be
yet another service to maintain in parallel to what we already have,
then it is hard for us to see whether maintenance costs are really lower
then the added value.

Those of us who operate top-BDII will eagerly replace it with something more effective but a new additional layer is something we are not convinced.