DAT40i and ARCserve - no compression?

[Steve Leaman]

I've been having an ongoing issue with compressing data with a DAT40i.

Some background info: I'm using ARCserve products (I've tried both version 6.61 as well as Brightstor's ARCserve v9.0) and am backing up a variety of uncompressed files - from CAD files (a small portion) to loads of MS Office docs. This is on a variety of OS's (I've tried SBS 4.5, Win2K Workstation, and Win2k Server), with a variety of hardware (switching everything from cables to SCSI cards to motherboards - I've tried custom builds as well as OEM builds from Dell and Compaq). The tape drive itself remained constant.

I then got an identical HP40i DAT drive and tested it in the same configurations - it provided exactly the same results.

In every case, I am unable to get the application I use to backup files to show any level of compression. I am able to get just shy of 20GB of data on a tape before it asks for a new one - to me that means I'm only getting the native format with no compression.

Since the combinations of software I'm using do not seem to be able to report a compression ratio, I tried backing up 21 GB of data. This resulted in the first tape running out of room and then asking for a new tape. I then tried using NT backup (set to take advantage of hardware compression) to do the same thing with the same data, and it too asked for a second tape.

I then tried to troubleshoot the problem with HP's Library and Tape Tools - this tool showed a compression capability of 2.23 to 1 - the maximum possible (I've used this same tool in all the different environments I've worked this problem in, and it gives the same results each time).

So here's a few questions:

Has anyone else suffered this problem?

If so, what did you do to fix it?

Is there another app (freeware or shareware) that I can use to test compression on this? I'm specifically looking for something that reports compression ratios after backup (ARCserve is supposed to, but has so far been unable to with this DDS-4).

Thanks in advance...

[Eugeny Brychkov]

Steve,
try native NT backup (on windows OS). To be sure you can use option "HC:on"
Eugeny

[Steve Leaman]

Thanks for the quick reply on this.

I've already tried using Windows Backup on both NT and 2000, and both failed the 21 GB test, even when I selected Hardware Compression. Neither were able to give me the compression ratio.

[Eugeny Brychkov]

1. check drive's jumpers (at the bottom side of drive assembly). All should be on except 3rd is off;
2. try zipping data you're backing up (if possible). Just to make sure that data will zip in less than 20GB.
I remember there were cases when software was disabling compression. I'm not sure, but remember turning jumper 4 off should solve issue (this jumper turns off software ability to turn off compression). I may not be correct, but it's worth to try
Eugeny

[Marino Meloni_1]

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=lpg50460#P180_15521

in this document there is a link to a programm called makedata, it will generate lots of files with know conpression, you can use it to troubleshoot.
It help me to find a problem related to the BU software,in fact when it started, it switched off the Hardware conpression.
Restarting the server (after a power down) without any backup service started let me understand the compression problem.

marino

[Glenn Weavind_1]

Steve: Demystifying compression: the hardware compression built into modern tape devices like DAT, DLT is good, carries no performance penalty, but is only a little more clever than redundant character suppression. Throw random data at this compression, and often the storage space required will actually increase by a few percent. All password protected files can be regarded as effectively pseudo-random, so won't compress. Many files will compress well, and overall a conservative figure to use is 1.6:1 -depends on o/s.
Compression by the backup product is provided mostly to minimise bandwidth for remote backups: it's costly (guzzles CPU cycles at remote end), is sometimes a little cleverer than hardware compression, but may reduce backup throughput, while delivering the benefit of reduced network traffic. When software compression is used, you would normally turn *off* hardware compression for reasons listed above - you're now firing pseudo-random data at the tape drive. Bottom line: you can get a quart into a pint pot, but not a gallon!