HP OpenVMS Systems

ask the wizard
Content starts here

Multisite Cluster and Shadowing Throughput?

» close window

The Question is:

Hello Wizard,
             we have 2 alphaserver GS80's using
2GB fibre channel adapters connecting to an enterprise virtual array for our
 disk storage. We are runing a split site cluster using FDDI as the intersite
 link. The EVA disks are shadowed via the FDDI link. We also have a
quorum node attached (3 node cluster)
When we remove one of the EVA's from the shadow set (Local processing via 2GB
 fibre) our application completes in 3hours 40 Minutes. However when we remount
 the other shadow set and run the same application it takes around 5 hours 40
 minutes. We have anal
yized FDDI and at
no time do we exceed 90% utiliation with shadowing switch on. We have conducted
 tests and
can run FDDI consistently at 97%.
We have an FDDI ring HALF DUPLEX - would making FDDI POINT to POINT at 100MB
 FULL duplex have a performance improvement. We have tested the FDDI
Adpaters but we are now at a loss as to why there is so much time loss between
 local and clustered performance.
Any Ideas?
Mark Caccavone
Man Financial IT

The Answer is :

  Please contact the support center for assistance in determining
  the particular I/O performance characteristics for the referenced
  communications configurations here.  That said, the OpenVMS Wizard
  would tend to expect that the relative network utilization levels
  mentioned here are well beyond the desirable utilitization levels
  (for the apparently-required performance) and that these levels
  verge on communications saturation.

answer written or last revised on ( 21-JAN-2003 )

» close window