Amanda-Users

Re: changed from hw comp to sw comp

2004-02-04 13:22:11
Subject: Re: changed from hw comp to sw comp
From: Paul Bijnens <paul.bijnens AT xplanation DOT com>
To: tobias.bluhm AT philips DOT com
Date: Wed, 04 Feb 2004 19:10:17 +0100
tobias.bluhm AT philips DOT com wrote:

I'm slightly baffled at the Tape Time though - it seems to have gone down. I expected it to remain about the same, half the data at half the tape write speed.

I'm posting my numbers & changes - comments?


Comparison between hardware compression & software compression

Amanda server = RH7.1, PIII 550 Mhz, 256M mem, 80G IDE holding disk
Tape Drive = AIT-2 scsi

I believe the native throughput of an AIT-2 scsi drive is somewhere
near 6000 K/sec.  (Verify your documentation.)
I have an AIT-1 with 3000 K/sec throughput.

Assuming you can indeed saturate the scsi-interface, then the limiting
factor of most tape drives is the speed at which the tape moves.
When using hardware compression, less bits need to be written to tape,
and you can need to feed more bytes to the interface. *If your data
compresses well*.
If your data does not compress at all, then many hardware compression
algorithms actually enlarge the data, and you'll notice a performance
slowdown instead, even below the advertised speed.

This difference in speed is tested with the program "amtapetype" to
detect if you have hardware compression enabled. Have a look in the source. (I wrote it.)




Typical results with large backup run using hardware compression:
...
Avg Tp Write Rate (k/s)  6541.9     7264.8     4388.6


Your full backups go over 6000 K/sec (the daily backups are slowed
down by stopping/starting the drive probably).



One of the first runs using software compression:
...
Avg Tp Write Rate (k/s)  5560.0     5919.3     4164.3


Your full backups are indeed near 6000 K/sec, what I would expect.

From the above numbers you can even estimate what the compression factor
is from the hardware compressor: 5919.3 / 7264.8 = about 80% (assuming
the tape moving is the only limiting factor -- if CPU, bus contention
etc limit the delivery of bytes to the tape drive, then you also better
off using software compression, because now the tapedrive can keep
streaming).



--
Paul Bijnens, Xplanation                            Tel  +32 16 397.511
Technologielaan 21 bus 2, B-3001 Leuven, BELGIUM    Fax  +32 16 397.512
http://www.xplanation.com/          email:  Paul.Bijnens AT xplanation DOT com
***********************************************************************
* I think I've got the hang of it now:  exit, ^D, ^C, ^\, ^Z, ^Q, F6, *
* quit,  ZZ, :q, :q!,  M-Z, ^X^C,  logoff, logout, close, bye,  /bye, *
* stop, end, F3, ~., ^]c, +++ ATH, disconnect, halt,  abort,  hangup, *
* PF4, F20, ^X^X, :D::D, KJOB, F14-f-e, F8-e,  kill -1 $$,  shutdown, *
* kill -9 1,  Alt-F4,  Ctrl-Alt-Del,  AltGr-NumLock,  Stop-A,  ...    *
* ...  "Are you sure?"  ...   YES   ...   Phew ...   I'm out          *
***********************************************************************


<Prev in Thread] Current Thread [Next in Thread>