ADSM-L

Re: A lot of questions concerning encryption and compression !!!

2002-08-08 10:16:46
Subject: Re: A lot of questions concerning encryption and compression !!!
From: Thomas Denier <Thomas.Denier AT MAIL.TJU DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Thu, 8 Aug 2002 10:19:18 -0400
> we have a problem with the capacity of dlt-tapes. Normally on one dlt-tape
> (35gb) you
> can store about 50 GB with compression. Our DLT-Library stores only about
> 30 GB !!!!
> So we searched for a compression-switch, but without success. The library
> says, that
> compression is on. We are using the tsm-encryption-feature for all files on
> every client.
> The tsm-client-release ist TSM V5.5.1 for Linux on most clients, our Server
> is TSM 4.1.3
> on Windows NT.
>
> Questions:
>
> 1. Does TSM (DES) encryption increase file size and if so, with which
> factor ??

I don't any hard data on this point, but I would not expect any significant
increase.

> 2. Does TSM (DES) encryption change the file-structure in a way that makes
> the dlt hardware compression
>     uneffective ??

Yes. All compression schemes exploit non-random characteristics of data,
such as differences in the frequencies of different characters or recurring
sequences of characters. All good encryption methods do a very effective
job of making data look like a random sequence of bytes, leaving nothing
for compression algorithms to exploit.

> 3. Is it in this situation useful to enable the tsm-ba-client-compression
> ???

The client compression would almost certainly reduce the size of the files
sent to the TSM server, at the cost of requiring more processor resources
during backup and restore operations.

> 4. If encryption and ba-client-compression is enabled => which of this
> options comes first ?!?!

Compression is done before encryption.

> 5. Do a lot of very small files shrink the tape capacity (because of gaps
> on the tape)????

I don't think small files cause the TSM server to generate lots of explicit
requests to write inter-block gaps. Helical scan tape drives might end up
with gaps because of the timing behavior of small file processing, but
DLT drives are not subject to that effect.

> 6. Does the volume-capacity shown on the tsm-server represents the real
> stored bytes or is it the
>     summarized value of all stored files given by the tsm-client-nodes
> (before or after encryption???) ?

For a full tape the volume capacity gives the number of bytes the TSM
server sent to the drive before encountering an end of file condition.
If client compression is used this may be considerably smaller than the
aggregate size of the client files whose backups are on the tape. For a
tape that is still filling the reported volume capacity is the estimated
capacity for the device class definition or the number of bytes written
so far, whichever is larger.