Do I have an Ultra160 Disk Bottleneck?

Hi all,

I've been trying to determine if I indeed have an Ultra160 disk bottleneck on our system. The system is a Netfinity 7100 running Win2k. I've got 3 RAID 5 arrays and 1 mirrored array. When running performance monitor during disk access slow downs, I notice that the disk queue length passes from 0 to more than 50 as soon as the total "disk byte/sec " counter reaches 40 MB/sec. It doesn't matter if the 40 MB/sec is reached on indivudual arrays or for all the arrays combined. Unless I am dead wrong, this is well below the theoretical 160 MB/sec that Ultra 160 is supposed to deliver.

Does that indicate I have a disk subsystem bottleneck? Is there something else I am missing or should check?

Thanks in advance.

Solution: Do I have an Ultra160 Disk Bottleneck?

That 160 number is rarely reached, and only if the disk cache is being used, which empties out in a matter of seconds. 80MB/sec is a more realistic number for maximum throughput on the outside cylinders with everything else optimized. A tool like HDTach will tell you what your array will deliver.

Popular Tags

Browse All Tags