This shows you the differences between two versions of the page.
— |
cluster:15 [2006/12/20 14:19] (current) |
||
---|---|---|---|
Line 1: | Line 1: | ||
+ | \\ | ||
+ | **[[cluster: | ||
+ | |||
+ | |||
+ | ====== Scali/ | ||
+ | |||
+ | * [[http:// | ||
+ | |||
+ | \\ | ||
+ | Like Platform/ | ||
+ | \\ | ||
+ | |||
+ | * heterogeous clusters (as in, manage the other clsuters on campus ...) | ||
+ | |||
+ | * " | ||
+ | |||
+ | * simultaneously deploys RPM installations (so you can perform entire disk image updates with the " | ||
+ | |||
+ | |||
+ | * parallel ssh & file copy support | ||
+ | |||
+ | |||
+ | * Change Management ... this is a biggie, for example, if you were to add a node: all nodes would need updating, this becomes automatic with change management, it'll auto detect what needs updating on other nodes | ||
+ | |||
+ | |||
+ | * Fault Handling and Root Cause Analysis ... also a biggie, know when something breaks before it happens | ||
+ | |||
+ | |||
+ | |||
+ | * Scali/ | ||
+ | |||
+ | |||
+ | * java/ | ||
+ | |||
+ | |||
+ | * it also supports PBS Pro, and MPI libraries and MPI/HA ... that is high availability for HA (reasoning goes like ... if jobs run 30 days and a single node fails and MPI is not HA then the entire job is aborted. So HA provides a pathway for atttempting to finish job while hardware underneath gets replaced). | ||
+ | |||
+ | |||
+ | * And lots more. | ||
+ | |||
+ | \\ | ||
+ | **[[cluster: | ||