Both Mark and Denys display their wealth of wisdom in matters pertaining to
backup, and I'm very glad they are on this list.

But I don't think either knows a cricket bat from a marsupials pouch :)

Cricket, Denys, is a splendid game played all over Great Britain and it's
former Colonies and Protectorates, except, it seems, in the USA, where the
game mutated into Baseball, now arguably the most popular bat and ball game
in the civilised world.
Cricket games can last up to 5 days, and then may end in a draw.

I do love your description of that disquieting state that tape drives go
into - shoeshine mode.

Regards

Neil



-----Original Message-----
From: Denys Beauchemin [mailto:[log in to unmask]]
Sent: Saturday, January 15, 2000 6:09 PM
To: [log in to unmask]
Subject: Re: testing of backup performance


I do not know what is more interesting, watching crickets or watching a tape
drive.  :)

However, here are some further thoughts:  Neil gets 1.7MB/s.  That is a
shade
under the 2.0 MB/s that is maximum for a compressed DDS-3 drive.  That is
very
good.  The backup to $NULL reveals a capability of stripping the data off
the
disk drives at 3.6 MB/s.  In this case, using a DLT 7000 would be a mismatch
and you would get only a small increment in performance.  The DLT 7000 wants
5
MB/s uncompressed of 10 MB/s of compressible data.  Neil's 3000 cannot
supply
that throughput so the drive would probably go into shoeshine mode and just
not
perform up to expectation.  Perhaps a closer match would be a DLT4000 (1.5
-3
MB/s) or a DDS-4 (3-6MB/s).

Kind regards,

Denys. . .

Denys Beauchemin
HICOMP
(800) 323-8863  (281) 288-7438         Fax: (281) 355-6879
denys at hicomp.com                             www.hicomp.com


-----Original Message-----
From:   Neil Harvey [SMTP:[log in to unmask]]
Sent:   Friday, 14 January, 2000 11:44 PM
To:     [log in to unmask]
Subject:        Re: testing of backup performance

As usual, Denys' take on this backup thing drives me to do my own obsessive
backup benchmarking....

On our 967, we usually get around 1720K/S (according to the statistics
option on store).

Every day, we send about 43GB to a single DDS3 drive (with DDS3 Media in
it), and "hardware" compression is turned on just before the store begins.

The DDS3 spec is 12GB native, and "24GB" compressed - clearly we are getting
better compression than this estimate with standard MPE Databases and raw
data.

The job runs for just under 7 hours, and this is when we eat, sleep, chat,
watch cricket, and occasionally visit our families.

A quick test to $NULL reveals that the system is capable of pushing out the
data at 3589 K/S, so we have the potential to half our backup time, if only
DDS4 or DLT/x was infinitely faster than DDS3...

Just for interest.....

Regards

Neil



-----Original Message-----
From: Denys Beauchemin [mailto:[log in to unmask]]
Sent: Friday, January 14, 2000 9:49 PM
To: [log in to unmask]
Subject: Re: testing of backup performance


Your tests are very interesting.  They only show one side of the equation
however.  You should try to do the same test but send the backup to $NULL.
 There are three/four components to a backup.  If the backup is to a local
device, the three components are: the disk subsystem, the software and the
backup device.  If the backup is to a remote device, the fourth component is
the network itself.

snip....