MIDRANGE dot COM Mailing List Archive



Home » MIDRANGE-L » February 1998

Re: Upgrade Options



fixed

At 03:46 PM 2/9/98 -0500, Gary Feinstein wrote:
>I am currently running a 510 with 512Mg of memory, 60 Gig DASD and 2143
>processor.  At most, I have about 50 users on at the same time.  Lately,
>some of my power users are running a number of resource intensive batch
>jobs at the same time.  My quetion is:  What is the best was to upgrade
>in order to get the most bang for the bubk.  Do I throw in another
>processor (bringing me to 2 out of a total possible 4).  Or do I throw
>more memory into the system.  I have heard that RISC is really greedy
>when it comes to memory.  How can I test my system to see which is the
>best option?  

Gary,

Run wrksyssts and look at your non-db fault rates. On my CISC system, I try to 
keep them under 13 (lower for the system pool - around 1-2). RISC will be 
somewhat higher, but I'm not sure what the guideline should be. I'd guess about 
20. If you have decent swap rates, and no ineligible waits, you have enough 
memory. If your CPU usage is close to 100% most of the time, you are CPU bound. 
Also, take a look at wrkdsksts. Your usage should be well below 50%. Below 20% 
is better. If it is high, you may need more disk spindles to handle the 
transaction volume. The drives should also be below 70% full.

hth
Pete



Pete Hall
peteh@inwave.com 
http://www.inwave.com/~peteh/

+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to "MIDRANGE-L@midrange.com".
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---






Return to Archive home page | Return to MIDRANGE.COM home page

This mailing list archive is Copyright 1997-2014 by MIDRANGE dot COM and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available here. If you have questions about this, please contact