Subject: | |
From: | |
Reply To: | |
Date: | Mon, 4 May 1998 09:44:29 -0700 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
In article <[log in to unmask]>, Lars says...
>
>At 11:23 28.04.1998 +1200, you wrote:
>>Since I switched most of my connections from serial to network (Reflection
>>using NS/VT) response has take a serious nose dive! Any suggestions where I
>>should start to look to resolve the problem, I didn't see anything in NMMGR
>>that looked very helpful. This is on a 918, 5.5 pp0.
>
>Do only NS/VT users suffer poor response times or DTC users or console, too?
>
>As far as I understand, NS/VT sessions need more resources on the 3000 than
>DTC sessions, for example an additional process (VTSERVER) per session. If
>your 3000 was already close to being "maxed out" (eg at memory) then these
>additional processes might have have shifted it into the "knee" of the MPE
>typical performance curve...
>
>Just an idea, in case the network resourse analysis won't give any clues...
>A performance tool like Glance or SOS should be helpful to invest the issue.
>
>Lars (guessing only for myself, not for HP, as usual)
I remember discovering that vtserver uses about as much cpu as the Quick
screens
in the application I looked at. So if most of your cpu time was being used
mostly for screens and not jobs NS/VT could increase its load substantially.
|
|
|