ADSM-L

Re: VM or MVS Server: Which Way to Go

1993-11-11 14:54:37
Subject: Re: VM or MVS Server: Which Way to Go
From: Leonard Boyle <SNOLEN AT VM.SAS DOT COM>
Date: Thu, 11 Nov 1993 14:54:37 EST
On Thu, 11 Nov 1993 08:52:38 PST Paul L. Bradshaw said:
>The MVS and VM servers both have the same basic ADSM function, so you will
>not be able to use that part as a determining factor.  Factors you should look
>at are:
>
>1.  Network connectivity
>2.  Support personnel skills
>3.  System capacity
>4.  Performance
>
>The MVS server has slightly better performance than the VM server due to some
>better Disk IO capabilities on MVS than on VM (10% - 20% better performance on
>disk related activities).  Other than that, the reasons to decide are based
>on network bandwidth to the processor, capacity (disk/tape/CPU) of the
>processor, and the comfort level of support personnel to run it.
>
>
>Paul Bradshaw

Paul 10%-20% seems to be a big difference in performance unless there is
something other than Disk IO capabilites that affects performance.

Is Disk IO performance directly related to total thru-put or
just the difference between the start time and end time for a single
backup/restore, for one workstation? Does that mean that MVS could
back up n workstations in a given time and VM would do n-m workstations,
where m is 10-20% of n?

How would VM mini-disk caching or having multiple mini-disks for the
catalog, log and storage pools affect performance? Is the 10%-20% using
the best of VM and the best of MVS services?


Assuming equal CPU size, equal dasd units, and using TCPIP is there
any other major differences between MVS and VM? Would MVS use more or less
CPU services for the ADSM and TCPIP services for the same work? For this
question assume that one is backing up several workstations at once.
The Red book on ADSM only gave performance numbers for the backup of
one workstation. I assume that most sites would be doing more than one
at a time.

How does working on more than one request at a time affect the performance
and thru-put of the ADSM server?
I would assume that the elapsed time would increase. But would the
measured amount of hardware resource per request be affected?

Can MVS utilize a multi-engine CPU better than VM?

Can a fault in the ADSM server shutdown all of the MVS TCPIP services?
<Prev in Thread] Current Thread [Next in Thread>