It depends on how many concurrent users you have for the device and the pattern of usage, but unless you have a very unusual environment, without PAV you will have a number of drives that regularly have I/O that encounter queueing delays waiting for access to the UCB, and these will see significant benefit from aliases. If you have a mix of short transfers and very large transfers (e.g., DB2 tables), even one alias can make a incredible difference on the consistency of short transfer response times. You will see a lesser improvement with increasing number of aliases until you reach enough so that no concurrent user for the device ever has to queue for access to a UCB address. Depending on your environment, even as low as 3 - 5 aliases for a 3390-3 may be adequate to give outstanding performance. In other environments, it may make more sense to let WLM handle PAV assignments.

McNeely, Tu wrote:
I would like to know if there are any significant improvement in
performance when using PAV?
And in your experience, how many aliases does it take per CU/device to
make a substantial difference?
...


-- Joel C. Ewing, Fort Smith, AR [EMAIL PROTECTED]

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to