SunCluster File System Performance (UFS/PxFS)

From: Martin Preßlaber (presslaber@ips.at)
Date: Mon Oct 22 2007 - 15:20:24 EDT


hi!

i've read threw some articles about the Cluster FS Performance and how to
increase it, but still i am missing one simple answerer...

Does it help to increase the numbers of interconnects?

i found in the blueprint from Ellard Roush, 817-1593 "Cluster File System
Performance", that the client side cache has to make round trips across
the interconnect, PxFS checkpoints cross the interconnect and the
acknowledgements go threw... I found in an interconnect description that
data is striped across all available interconnects in a round-robin
fashion. So looking to the prices of gigabit copper ethernet cards
nowadays, it's no matter about money to increase the number of
interconnects like mr. roush mentioned in his blueprint in 2003... i
found a hint in my cluster-course training book "you can have more than
two that would be a performance benefit in certain circumstances ( <--
??? ), because global data access traffic is striped across all of the
transports"
so well, easy question, but i couldn't find any answerers.. does it
actually help to increase the throughput by adding additionally
interfaces?

thanks in advance,
regards,
- Martin Presslaber
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers



This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:42:27 EDT