ADSM-L

Re: Gig utilization during backups

2006-09-08 16:16:08
Subject: Re: Gig utilization during backups
From: "Darby, Mark" <Mark.Darby AT HQ.DOE DOT GOV>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 8 Sep 2006 16:14:31 -0400
We have a z/OS.e 1.6 OS running in an LPAR on a zSeries (z890-270) using
a single (shared CP) engine, single (shared) gigabit Ethernet (OSA
Express) adapter port.

We're running a 31-bit TSM server (that's all there is for us thus far)
- 5.2.6.2.  I mention this because this TSM server is running in a
"process" limited to 2GB of virtual storage, so the real storage size of
the LPAR in which it is running is somewhat superfluous, except to say
that the OS is not paging/swapping at all.

Our actual measured consumption statistics vacillate between in around
200-500 mbps, depending on the time period in question.  I would say
that we see around 350-400mbps (median) READ rates for many intervals
each night with peaks of 650-675mbps for several intervals every week.

We back up about 1100 unique file systems from about 400 nodes every
evening in a 6PM-6AM window and average about 1.2 TBs of backed up data
per day (24 hrs.).  The 6AM-6PM load is minimal, but I would guestimate
that the actual 6PM-6AM data volume is about 1.1 TB, therefore.  We have
a large reserve capacity that could probably sustain 4-5 TBs per 6PM-6AM
period, if we were to provide enough DASD to absorb that much data.

We have a maxsession set at 200 with maxschedsession set to 99%.

The rates I've quoted above are actually "measured averages" of 10
minute intervals.  In other words, I collect the interval statistics
from our drivers which give us counters at 10-minute intervals.  The
rates I have quoted are determined by dividing the number of bytes READ
in the interval by the number of seconds in the interval, so the actual
peaks and valleys in these statistics are unknown.  Then, of course, I
convert those numbers to mbps (millions of bits per second) in keeping
with what I believe are normal network measurement terms (using 8-bits
per byte in the calculation).

Regards,
Mark

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Willeat, Todd
Sent: Friday, September 08, 2006 2:24 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: Gig utilization during backups


We see 115MB/s - 120MB/s on a single GigE. We have 2 boxes that are
p570s with 4 CPUs and 8GB RAM. Each runs 3 TSM 5.3x instances on AIX
5.3...

               _______________________________________
              /            Todd A. Willeat            \
             |          Storage Administrator          |
             |  Sisters of Mercy Health System - MISD  |
             |            3637 S. Geyer Rd.            |
           _ |           St. Louis, MO 63127           | _
          / )|  (314) 364-3110 / (314) 364-3512 (fax)  |( \
         / / |          willt7 AT stlo.mercy DOT net          | \ \
       _( (_ |          http://www.mercy.net/          | _) )_
      (((\ \> \/->___________<-=-=-=-=->___________<-\/ </ /)))
      (\\\\ \_/ /                                   \ \_/ ////)
       \       /                                     \       /
        \    _/                                       \_    /
        /   /                                           \   \
       /___/                                             \___\


This e-mail contains information which (a) may be PROPRIETARY IN NATURE
OR OTHERWISE PROTECTED BY LAW FROM DISCLOSURE, and (b) is intended only
for the use of the addressee (s) named above. If you are not the
addressee, or the person responsible for delivering this to the
addressee (s), you are hereby notified that reading, copying or
distributing this e-mail is prohibited. If you have received this e-mail
in error, please contact the sender immediately.


-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Gill, Geoffrey L.
Sent: Friday, September 08, 2006 11:50 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: [ADSM-L] Gig utilization during backups

I'd like to get some input on what others see on their port utilization
during heavy network traffic. With 350+ systems backing up to a single
TSM server I'm curious as to why we see only about 50% utilization on
the GIG port with max scheduled sessions set to 95. Sessions run
extremely long compared to when they were on the old system.



What have others seen their systems and what do you have set for max
scheduled sessions?



I still feel like this system, which has 4 CPU's is not nearly as fast
at moving data than my 2 CPU system was and the setup is very similar.



AIX 5.3 - TSM 5.3.3.0 - 4 CPU's - 4GB memory - Gig interface





Thanks,



Geoff Gill

TSM Administrator

PeopleSoft Sr. Systems Administrator

SAIC M/S-G1b

(858)826-4062

Email:  <mailto:geoffrey.l.gill AT saic DOT com> geoffrey.l.gill AT saic DOT com

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