Minutes, workgroup meeting 23 May 2006

We have so far provided a survey of Unix/Linux users, Unix user programs and frequency of use among various programs, and Unix servers and the tasks they perform. We now have to plan the future IT structure.

This group has so far been most reluctant to the idea of migrating to Linux (background information here?). However, the cost and technical difficulties related to migration might have been slightly exaggerated; Seisan compiles under Linux, and Seisnet is based on Expect (script language) and some Fortran, so is does seem possible to port SeisNet to Linux. The Earthworm (more info from Orfeus) seismic data collection system that is currently being evaluated is based on SUN/Solaris & Windows, but there are mailing list entries discussing Linux porting (initiated by British Geological Survey (BGS), who considers updating Earthworm servers - coordination possible, we are cooperating with BGS ..?)

These options exists:

1. Seismology group continues to run on Sun Sparc/Solaris, either on existing hardware or new server box (Ø.N.: We should avoid administering two operating systems, Solaris and Linux, locally).
2. As A) but in addition given instrumentation system status, placed behind firewall provided & managed by central IT.
3. Moving to Solaris servers that central IT will set up (available from Q1 2007?)

GMT and Seismic Unix, two programs getting high scores in the user survey can (eventually) be part of the Linux client system. More general purpose programs (like MatLab) will of course also be included.

There is a 'generation transition' among scientific staff that will probably result in increased demand for local computation resources. Also CIPR and IFG can be considered as a unit in this regard. These computation intensive fields identified:

1. Seismic processing
   • Industry seems to prefer Linux. Clusters are common.
   • Three programs will be used: Seismic Unix, Promax, GeoCluster (CGG)
2. Basin modeling. R.H. has been airing thoughts about local linux cluster, medium size (128 nodes), with shared funding. B.T.S. says we can have this cluster locally; central IT would, however, probably not consider managing these machines due to close proximity to BCCS turf.
3. Reservoir modeling
4. Geodynamic modeling

No cookies served in this meeting. Don't forget next time!

Where are results from User survey wine quiz? (Will soon be put on web page.)