|
HP OpenVMS Systemsask the wizard |
|
The Question is: In a recent answer, you stated In this particular configuration, you will particularly want to consider the bandwidth of the CI adapters and the activity on the CI buses. Unless your present storage configuration is severely I/O bound and you have similar storage technologies with those of your planned upgrade, you will likely see improvements from any storage upgrade. (Barring an existing I/O performance limit -- CI bandwidth, for instance -- or an existing non-I/O performance limit -- eg: CPU performance -- that is not specifically related to the disk or RAID controller performance.) Thus the use of the larger disks may be more cost-effective. This begs the question: How does one measure the bandwidth used on one's CI adapter? MON SCS? MON DISK /ITEM=QUEUE might show an IO bottleneck but will not reveal if the CI is the cause or the disks living beyond the CI. Does one need to aggregate the KB/S or RQ/S fields of the VTDPY screen inside each HSJ controller? The Answer is : The kilobyte mapping rates are the aggregate rates for each SCS path. These values are visible in MONITOR SCS and SHOW CLUSTER/CONTINUOUS. Various performance monitoring tools can also be used. In conjunction with knowledge of the CI cabling, the aggregate rate for each star can then be determined. Aggregate CI bandwidth is roughly 17.5 megabytes per second, spread across two paths -- recent CI adapters will operate on both paths in parallel, while older CI adapters are restricted to use of one path at a time.
|