Notes from 10/22/99 CEASE meeting Attending were Abby, Bob Basch, Bill Cattey, and Naomi Agenda ------ 1. Data since last time, especially responses from faculty 2. Grid of cluster renewal scenarios vs stakeholder impact 3. Phaseout scenarios for Dept Athenized SGI's 4. Helping courses transition off Athena SGI's 5. next steps for CEASE Naomi presented a summary of faculty responses. The major impact would be those departments (Chemistry, Chem Eng, Mat Sci) that depend on MSI software. There are also the EECS compiler course whose students test their compilers on MIPS machines, and a few courses using OpenInventor. Bill has received a reply from the SGI regional service manager saying that he'd try to help with our latest bug, but no specifics. SGI still hasn't sent us a loaner of a 9 GB O2 so that we can qualify it for Athena now that they're no longer selling 4 GB systems. Recommendations that came out of our discussion: ----------------------------------------------- Do not buy any further SGI O2s. Also ... 1. Do not automatically remove the 44 FY97 Indys in the public clusters in the summer of 2000, which would be the end of their normal 4 year life cycle. If we want to deploy a new platform such as Pismere for no more than 22 of these seats, we might do so, either during Summer 2000 or during IAP 2001. 2. Work with Chem Eng, Chemistry, and Mat Sci to launch a project to meet their MSI needs in the future. This might mean moving toward the architecture of an MSI server with Linux or NT clients, something that MSI appears to be planning. Bill proposed that a Discovery Project be launched for this, to include Integration. 3. In Summer 2001, remove ALL remaining Indy's from clusters. This means that in Sept 2001 there will be 49 O2s in the clusters. (There are currently 49 O2s and 44 Indys, amounting to ~1/4 of all public machines). 4. Remove the 4 GB O2 machine as an option to purchase from our web pages. Wordsmith caveats with owls if and when we do receive a 9 GB O2 to qualify as an Athena machine, so that people who want to purchase O2s from now on are adequately warned that they may not have a full 4 year lifetime. 5. Send out a letter with the June 2000 billing saying that the end of life for departmental Indys will be June 30 2001. 6. For the 6.035 compiler course that uses SGIs because of their MIPS processors, we can either offer them the Indys when we remove them from public clusters, or else help them find a MIPS emulator that we can put into a locker for use on the Sun workstations. 7. For people who use OpenInventor software on the SGI - there are relatively few such people and relatively few students/courses affected. We can help them transition to non-Athenized SGI's in departmental clusters (e.g. Ocean Engineering, which has 3 IS-granted SGI and only about 10-12 majors in a relevant course). 8. Homegrown software - One person mentioned the desire to have help porting the software to Java. We can throw ACS $ at this. As for other stuff, where there is no longer any source code, it will need to suffer normal code rot. 9. Naomi still needs to talk with Seth Teller to find out what his graphics needs are. She's meeting with him on Monday. 10. Alias-Wavefront - tfitz currently maintains this for SGI only (alexp manages liserver); it also exists for Sun, but this may be a recent development of which he's not aware. Naomi will speak with him; transition proposal for these users to follow. 11. Once the O2s that we have just purchased expire (Summer 2003 = FY2004), we will follow the rule that (end of life of private machines as Athenized systems) = (end of similar machines in the public clusters), with plenty of advance warning. Abby will start to draft a report with recommendations and will circulate it to the team. Whether or not we meet next week will depend upon how far she's gotten.