Shane wrote:
On Thu, 2006-09-28 at 17:53 +0000, Manuel Tabares Solórzano wrote:
We have a 9672 r16 with 5 lpars (four are os/390 2.10 and one is
vse)
We have experienced low response time in batch (workload).
My manager need to know: HOW MANY LPARS IS OPTIMUS ON ONE 9672 R16
?
Optimal - one. Else as few as possible.
Adding LPARs is expensive - especially on a uniprocessor.
With 5 (all busy) LPARs, basic maths says each is (on average)
nondispatched for 4 times as long as it's actually dispatched.
And each is guaranteed a non-dispatch time - probably 250 ms for the
R16. So *every* LPARs gets dispatched (and interrupted) every 1/4 of a
second.
Just ugly with that many.
Get it down to around two and things sound a whole lot better.
The above is correct only if LPAR weights are equal. It is seldom true.
--
Radoslaw Skorupka
Lodz, Poland
----------------------------------------------------------------------
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
