I can support Terry's statement. I have tested with high performance file
servers of two well-known manufacturers and both dropped to below 500
Kb/sec, using LTO, with an NDMP backup of a directory with 125
subdirectories with 1000 files of 400 bytes in size (125 directories x 1000
= 125000 files in total).
The reason for this performance drop has not been researched as it was a
special test case.
Bye
Ernie
-----Original Message-----
From: Terry Lemons [mailto:lemons_terry AT EMC DOT COM]
Sent: Thursday, November 21, 2002 10:40 PM
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Subject: Re: [Networker] throughput stats?
Throughput is dependent on the capability of the platform (as Bob's results
showed), but also on the data being backed up. If you back up zillions of
little files, your transfer rates will be very slow, no matter what platform
you're backing up on. On the other hand, large, compressible files deliver
startlingly-good results.
tl
-----Original Message-----
From: Bob Schuknecht [mailto:Bob_Schuknecht AT HILTON DOT COM]
Sent: Thursday, November 21, 2002 4:35 PM
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Subject: Re: [Networker] throughput stats?
Mark,
We see average transfer rates in the 12MB/s range from our more robust Unix
clients.
Other clients average 6-8MB/s. The real dogs (typically Windows clients)
generally deliver
2-4MB/s. We do have one Windows client that can sustain 10MB/s but nothing
else is
running on it at the time and it's a storage node.
ADIC Scalar 1000 AIT-2 lib
Networker 6.0.2
Servers Solaris 7 and W2k
-Bob Schuknecht
> -----Original Message-----
> From: "Legato NetWorker discussion"
> <NETWORKER AT LISTMAIL.TEMPLE DOT EDU>@INTERNET@HHC On Behalf Of
> marc.levitan AT PFPC DOT COM
> Sent: Thursday, November 21, 2002 10:17 AM
> To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
> Subject: [Networker] throughput stats?
>
> Hi all,
>
> Just wondering if my data throughput stats are up to par?
>
> Environment:
> networker 6.1
> server / client = solaris 8
> SAN using DDS
> adic scalar 100 LTO library
>
> is 24GB/hour good, bad, just alright?
>
> Start: Thu Nov 21 10:20:03 2002
> End: Thu Nov 21 10:43:10 2002
> Elapsed: 0 hour(s), 23 minutes
>
> Level Size in Gb
> ----- ------------
> full 9.31 Gb
> ----- ------------
> Total 9.31 Gb
>
> Throughput is 6.87 Mb/sec
> or 24.17 Gb/hour
>
> Thanks,
> Marc Levitan
> Storage Management Specialist
> PFPC Global Funds Services
>
> --
> Note: To sign off this list, send a "signoff" command via email
> to listserv AT listmail.temple DOT edu or visit the list's Web site at
> http://listmail.temple.edu/archives/networker.html where you can
> also view and post messages to the list.
> =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
>
--
Note: To sign off this list, send a "signoff" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
--
Note: To sign off this list, send a "signoff" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
--
Note: To sign off this list, send a "signoff" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
|
