In the Physics Division there are several groups involved in different types of research programs, detector R&D, big collaboration experiments short and long term, Astrophysics projects, and all with very different needs. A feature of these large collaborations is that the local LBL group often needs to follow the computing model determined by the host lab (FERMILAB, SLAC, CERN) rather than the LBNL computing model, for example for supported platforms, software tools, and so on.

The systems currently in use span from UNIX workstations serving many users through X-terminals or PC desktops used as a terminal clusters of PC desktops where the data are stored locally, and still some VMS systems. There is also a substantial use of the LBNL-NERSC PDSF cluster. In addition, there are several Windows 98, and WNT PC systems used for word processing, graphics, and data acquisition.

Computers for data acquisition are not considered in this document but they constitute not a negligible fraction of the computers used in the physics division. They are mostly PC’s (WNT) and in some cases real time UNIX processors. They are the crucial for detector development and production for large projects. Maintenance and backup are crucial for these machines, and they should not be completely ignored in the computing needs. So far, maintenance for PC’s has not been fully satisfactory while very expensive. This is an area that needs improvement.

Many people in the division spend large amounts of time doing mobile computing away from the lab, typically using laptop machines running Windows or LINUX. Of course, high bandwidth access is critical, but with support for DSL and cable modems, this situation is much improved. Wireless solutions for mobile connectivity also start to be interesting, although their bandwidth is still quite limited. Mobile users need to be able to see a consistent computing environment, so issues such as remote printing, and remote backup become important. At present, everyone is left pretty much on their own to solve these problems. The standardization of IMAP4 Email support has been a big help, and in the future tools such as VPN that help users be closely linked to LBL computing resources when they are off-site would be helpful. Security for these off-site users is also a big issue, and uniform support for an encrypted environment would be useful (SSH is a start but does not go far enough). Proposals for standard solutions in this area are of great interest to our division.

Just recently, due to budget cuts, representatives of each group started to discuss how to minimize the cost of computing with of course no impact on the research program. The main immediate goal for the physics division computing is to reduce the cost for the maintenance of UNIX workstations.

This can be achieved by replacing multiple workstations with fewer centralized servers for which maintenance is required and have more uniform desktop systems. A couple of groups already succeeded in this process by replacing expensive workstations with PC clusters, running LINUX and maintained by an internal support person. Also backup services, so far very expensive, have been replaced by internal support. It remains to be seen how well we can cope with the difficult problem of guaranteeing the integrity of everyone’s file base wherever they are doing their computing.

Several groups are presently working on developing improved computing environments for their near future needs. We also hope that the uniformity of desktops might be a path to evade heavy requirements from the security police impacting the research work directly. DAQ machines are a big problem in this respect.

Participating in large international collaboration projects involves a great usage of video-conferencing and exchange of documents through the web, hence heavily depends on a high-level network system.

Cost-effective videoconferencing is crucial to keep in touch with such projects. Some solution for remote (e.g. home DSL) video-conferencing would be quite useful. In general, telecommunication is an area that needs to be supported at the highest performance levels available, and the development of cost-effective solutions is greatly appreciated.