ADSM-L

Re: Netware 4.1 and File Compression

1996-11-27 12:55:30
Subject: Re: Netware 4.1 and File Compression
From: Francisco Reyes <reyes01 AT IBM DOT NET>
Date: Wed, 27 Nov 1996 12:55:30 -0500
On Wed, 27 Nov 1996 17:42:56 +1300, Dianne Sharp wrote:

>We have 3 Netware 4.1 servers which support data compression.
>In the ADSM Client option file, I have not specified COMPRESSION as an
>option, therefore I understand it should default to NO.
>The Compression option for the Nodes is Client determined.
>
>However, it appears that the data is sent to the ADSM Server (V1 on MVS)
>in compressed format because the amount of data checks out with the
>compressed amount on the Netware servers.

Check to see if you have any scheduled operation which indicates
-compress=yes. It seems that compression is a toggle. I had the same
situation in which I had one directory I wanted it to be compresses,
situation in which I had one directory I wanted it to be compresses,
but after that directory was done then the client continued using
compression.

>THE PROBLEM :
>It appears that when the ADSM backups are running, the Netware server
>CPU utilisation goes right up because, according to our Netware "gurus",
>it is uncompressing files for ADSM to read.
>Does this mean that they are uncompressed by Netware, then compressed
>again by ADSM?

Compression and Decompresssion is done by the netware client. Or to
be more general compression is done by the clients. This is actually
a good thing since if you have many different computers running at
the same time the ADSM server will have less work since it is getting
less data to work with. This is actually distributing the compression
to each client.

I use compression on my netware servers and it does not increase
utilization that much. Find out how much memory you have on those
netware servers. If they don't have enough memory then that could be
your problem. Moreover, did you ever use the "processor" option on
the dsm.opt file? If you did and you placed a number greater than 1
then reduce it. If memory is low then use "slowincremental" in your
dsm.opt.
<Prev in Thread] Current Thread [Next in Thread>