NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
>From: Mark Tucker <address@hidden> >Organization: Lyndon State College >Keywords: 199905111806.MAA13803 X-Server software Hi Mark- >I've been trying to plan some of the changes and upgrades we are going >to be making to the meteorology lab this summer. One of the possible >changes I've been considering is possibly running our applications >remotely via X from a server to PC workstations. Bill Fingerhut told me >that he had spoken (e-mailed) with you about this some. Is this >configuration something that Unidata would support? We really don't support this because we do not use PC X-Window software in our support group and have no easy way of testing if there are problems. Plus, even if we did use PC X-Windows software, the version we use might not be what you use. >If you think there >would be any reluctance or difficulty for Unidata in supporting this >type of setup I will not move in that direction. We can help you as much as we can without being able to test things locally. However, we do not officially support this configuration. >Obviously, I do not want to create something that will perform poorly or >be a headache to support. What conserns come to mind about this setup? >I hope to give this idea an evaluation soon to see if the server could >handle even part of the load. I am thinking of running this on something >like a dual P450 with about 1Gig of RAM. The biggest problem with using X software is that it will create a tremendous load on your server in terms of memory and disk access and the other systems in terms of network traffic. If you were going to do this you should have a machine designed for such serving like a Sun server or any other vendor server. I would advise not using the same machine for your LDM or you might see data loss if the load is great and the queue cannot be processed fast enough. There are several issues that need to be addressed. First, each machine will have to log in to a different account on the server to run McIDAS. You _CANNOT_ run McIDAS from the same account multiple times on the same machine, without creating problems. Second, you will need to make sure that the default display for the X server is 8-bit because both McIDAS and GEMPAK are designed for 8-bit displays. Bill mentioned that you were planning a test by running displays of 10 sat images at the same time. This will only test the display load, not the load of things like calculations. Try having all sessions run one of Bill's BATCH files that does a lot of calculations or some of the cross sections. You should also try running GARP from all the machines and continuously load loops and run grid calculations. GARP is a huge resource hog and that may be a better test. Again, the display portion is only a small part of the equation. >If you know of any other sites that are running thier applications in this >manner please let me know as I'd like to make a more educated decision. Plymouth State was using X server software on PC's in their labs at one time. You could contact Jim Koermer (address@hidden). I think University of Arizona was also running in this configuration also. You could send out a note to the community mailing list and ask if others are doing this and the pros and cons of such a setup. Don ************************************************************* Don Murray UCAR Unidata Program address@hidden P.O. Box 3000 (303) 497-8628 Boulder, CO 80307 ************************************************************* Unidata WWW Server http://www.unidata.ucar.edu/ McIDAS Demonstration Machine http://mcdemo.unidata.ucar.edu/ *************************************************************