Dear All
I have asked for an official statement from the dCache SRM
developers/consortium. These TB-SUPPORT discussions are useful but any
conclusions reached would benefit from being based on a confirmed
position from the dCache SRM provider(s).
As you are probably aware there are very few suitable SRM
implementations that sites can currently use. dCache SRM and the CERN
developed Disk Pool Manager (DPM) are really our only options. DPM is
still something of an unknown while dCache is being used in production.
I understand that there is concern about access to dCache source code
that needs to be addressed, but these discussions are also introducing a
new set of questions about what sites want and expect. Ideally everyone
would have a full knowledge of everything they deploy but this is not
practical so we as a region need to decide what we can accept - as John
suggested is it enough to have UK validation. Alternatively are all
sites expecting to have the ability to view source code so that site
bespoke solutions can be developed?
These questions will be raised during the GridPP13 deployment
discussions and you are welcome to join the debate:
http://www.gridpp.ac.uk/gridpp13/programme.html
Kind regards,
Jeremy.
-----Original Message-----
From: Testbed Support for GridPP member institutes
[mailto:[log in to unmask]] On Behalf Of Andrew McNab
Sent: 28 June 2005 10:28
To: [log in to unmask]
Subject: Re: d-cache software license?
Kostas Georgiou wrote:
> I *do* try to find holes at d-cache (or any other software) and if
> i find a security hole in a closed source software my only option
> is to disable it. On the other hand if it's open source i can try
> to fix the problem. Which software do you think is going to be better
> a few years down the line?
And furthermore, I think research projects which depend on particular
closed source software are built on sand: we have no way of tinkering
with the stuff to make it do what we want, and the whole point of
research is to do things that no one has thought of doing before, so
the "we could do X if Y had this little extra feature" comes up all
the time.
Take Objectivity vs Oracle vs MySQL as an example. Objectivity turned
into a dead end because it was a big ugly closed source solution that
we had no control over. Oracle is a similar sort of beast (although
better maintained because they're such a large operation of course)
but if we suddenly found we had to add some Grid stuff to an SQL
database (writing access policies in terms of Grid credentials for
instance) then we could use MySQL instead and hack our wish-list into
it and feed the changes back to them (that's what Yahoo do when they
use MySQL.)
It sounds like d-cache has become another bait-and-switch "oh well,
maybe we'll keep the source closed after all" game.
Cheers
Andrew
|