 |
 |
Thanks (I noticed the location). But I think John's concerns may still be valid in the long term. Atlas sites (and the other experiements, for that matter) will, in the long term, need to upgrade the OS many many times over the life of the experiment, and each "grid" site will have their own "required" levels of software (for functional, security, operational, staffing reasons). What is the plan (is there a plan?) to quickly certify each analysis site's level of software? BaBar has a method that works (more or less) because there are only a few sites involved with the "heavy lifting" of the analysis. With the "grid", it seems it will be (essentially) impossible to dictate a small number of variations, and with (for example) each university deciding on their own distro (I hear Stanford has chosen Ubuntu), and their own software levels, how will the experiment "certify" those sites unique distinctiveness (and add it to the Atlas "grid")? Gary (Troublemaker) Buhrmaster > -----Original Message----- > From: Young, Charles C. [mailto:[log in to unmask]] > Sent: Thursday, February 16, 2006 10:12 AM > To: Halperin, John H. > Cc: security; atlas-sccs-planning-l > Subject: RE: FW: [Usatlas-users-l] dual-core Opteron evaluation server > > This is a message from BNL about what they are doing that I > thought people may want to know about, and perhaps to > subscribe to their mailing list for similar items.