1 / 25

CFI 2004 Clusters @ UW

CFI 2004 Clusters @ UW. A quick overview with lots of time for Q&A and exploration. Q : What is a cluster? A : A group of machines that can work together to produce results. ?. ?. Characteristics. Group of machines Not necessarily homogeneous Common set of users

Download Presentation

CFI 2004 Clusters @ UW

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. CFI 2004 Clusters @ UW A quick overview with lots of time for Q&A and exploration.

  2. Q: What is a cluster? A: A group of machines that can work together to produce results.

  3. ?

  4. ?

  5. Characteristics • Group of machines • Not necessarily homogeneous • Common set of users • Some sort of batch system

  6. Why do I want one? Ideally your jobs are able to be parallelized and do not require allocation of large chunks of memory.

  7. When do I not want one? When you need large chunks of memory, or when you only need a single fast CPU

  8. What do we have? • marroo - 21 nodes • shiraz - 21 nodes • vidal - 16 nodes

  9. How do I get to them? Accounts - SSH client -

  10. Quick Hardware Specs • SunFire X4100 - dualCPU, dual core • 2.4GHz Opteron cores • 8GB RAM

  11. Filesystems • /home - big, shared • /scratch-net - big, shared • /scratch - local, not so big • Network filesystems on a NAS

  12. Backups There aren’t (yet?) any, so be careful

  13. Power shiraz and marroo have a UPS each backing head nodes and NAS vidal will be getting a UPS to back at least that also

  14. What can I do? They’re Linux boxes, so anything you could normally do on your workstation. Plus a bit more.

  15. Sun N1 Grid Engine Fairly simple setup of batch system, our implementation has only a single queue.

  16. N1GE QuickStart Only on vidal sge-root is /home/N1GE6 Cell name is “default” Important commands are qsub and qstat

  17. Commercial Software CFI grant has allowances for MatLab CPLEX is installed on vidal, arrange for licensing

  18. Free Software Of course, there are lots of packages available for Linux. We have installed R on vidal, along with some packages.

  19. Updates and Security Compute nodes as static as possible Head nodes receive security updates Head node reboots scheduled in advance

  20. Other Stuff Ganglia for rough usage data and trends Head nodes monitored for connectivity Mailing list!

  21. Hopefully this…

  22. … leads to this: Fun in the Sun

More Related