Networker

Re: [Networker] Drive speed?

2011-12-09 16:19:22
Subject: Re: [Networker] Drive speed?
From: George Sinclair <George.Sinclair AT NOAA DOT GOV>
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Date: Fri, 9 Dec 2011 16:18:04 -0500
On 2011-12-09 15:37, Eddie Albert wrote:
What is the purpose of tweaking the job for performance? Is the purpose
to backup asap to release your business application?


I was really just trying to review what sending a second save set to the same drive will achieve in terms of the one that's already running to that same drive but is going slowly. Apparently, sending additional save sets can not only up the speed of the drive, but in the process, it will also make the slow one more efficient since the drive is now streaming.

The slow save set is listed as an enumerated save set under the NSR client resource as: /pathname/blah-blah-blah. We usually get pretty good write speeds, but this one was running at a time when there was little else going on, so that may be why it was running so slowly since there was nothing else writing to the drive. Also, there could be a ton of inodes under there, fragmentation, etc. This is the first time that I've backed it up on that system, and that client is faster than the one where it previously lived. But then again, I'm usually not watching the write speeds since these normally run after hours. But when I then started another different save set (manually, from the client), the write speed on the drive ramped way up. It's not surprising, but I was mainly asking how that will affect the write speed on the original save set that was writing slowly to that same drive. I should have checked it using nmc since it will show you the Rate (KB/S) for each save set, but I didn't do that. I was just looking at the overall write speed for the drive.

What does the backup configuration look like?
Saveset=all
or
Saveset=C:\
Saveset=D:\
Saveset=E:\
Saveset=F:\
Saveset=G:\
Saveset=H:\
Saveset=I:\

If you can't tweak the saveset/performance consider changing out the
backup device to DataDomain or Avamar?

I'll have to check to see why it was still running when I came in this morning. It might be that the group didn't start that save set until very late, and maybe by then, most of the others were done so the drive slowed up. There are some paths that have a huge number of inodes, and while not terribly large, they can take a long time to back up - much larger than a similar sized file system with fewer files. Usually, the performance is acceptable as most of the data is running concurrently with other save sets, and the drives stream OK.


The answer to my first question, is the justification for my last
comment above.

Semper fidelis et paratus, /ALE


-----Original Message-----
From: EMC NetWorker discussion [mailto:NETWORKER AT LISTSERV.TEMPLE DOT EDU] On
Behalf Of Brian O'Neill
Sent: Friday, December 09, 2011 2:52 PM
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Subject: Re: [Networker] Drive speed?

On 12/9/2011 2:37 PM, George Sinclair wrote:
On 2011-12-09 14:07, Chiravuri, Sri, GDH Consulting/US wrote:
Excellent point. If this is the case - saving same client/saveset
twice
(in parallel) could also bounce the thruput up ?

Interesting theory. If it's true that both benefit - and, yes, I can
see
that - then what if you were backing up 50 GB of data for a single
stream that might otherwise take 45 minutes, let's just say. But, if
you
instead backed the same save set up twice (in parallel), could it
still
finish faster than 45 minutes? Hmmm ...

George

It really depends on the client and network performance, but I actually
think this won't work at all. The client couldn't keep up to the tape
drive speed for some reason - either the disk is slow, or the I/O
throughput of the system, or the network connection - all factors that
probably would not change no matter how many save streams are running
off that single box.

That, and I'm not sure how networker would react to trying to perform
the same backup at the same time.

You are much better off trying to schedule the slower systems to run in
parallel with other systems.

Also, even if you could do this, now you need to back up twice as much,
and could still take longer than the ideal speed.



Thx
Sri

-----Original Message-----
From: EMC NetWorker discussion [mailto:NETWORKER AT LISTSERV.TEMPLE DOT EDU]
On
Behalf Of Brian O'Neill
Sent: Friday, December 09, 2011 1:04 PM
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Subject: Re: [Networker] Drive speed?

The slow speed on the one stream is likely because your
client-to-server
data stream is below the minimum threshold of the tape drive to
prevent
it from having to write, pause, back up, spin back up to speed,
write,
pause, etc. So the total throughput you are seeing is low because of
the
time spent not actually writing data to the tape.

The second stream brings your client-to-server data throughput over
the
minimum of the drive, so the tape drive doesn't have to stop, rewind,
start again, lather, rinse, repeat.

Both streams should be benefitting - the first stream was penalized
by
flow control while the server waited for the tape drive. Now it
doesn't
need to.

On 12/9/2011 10:51 AM, George Sinclair wrote:
A basic question here on drive speed, but maybe not a simple answer
as

there are undoubtedly numerous variables involved.

Let's say you have an LTO-4 drive (SAS connection to the tape
library)

with a single stream (one save set) clocking in around 4-10 MB/sec,
coming in over the network. You then start another backup (also a
single
stream) from the same client to the same drive, and now it jumps up
to
70+ MB/sec, and remains at that speed until that second save set
completes, and then quiets down to 4-10 MB/sec again. I've seen this
happen with a number of other streams, too, wherein running just one
of them from that same client, concurrent with the already running
stream, cranks the speed up considerably, until it's done, at which
point the original stream is reported again to be running at the
same
slow pace.

We all know that a drive will come closer to performing optimally
when

you can keep it streaming, and you can do that by keeping its buffer
full. OK, so having more concurrent streams - up to a point - will
improve drive performance, BUT does it affect the speed at which the
slow stream runs?

In other words, when the reported write speed jumps up to 70+ MB/sec
because you're now sending another stream (possibly one that
compresses well), is the original stream (possibly one that does not
compress so
well) now increasing its write speed as a result? Or is it instead
the

case that while the drive is now functioning more optimally, and
writing more data per second, that first (slow) stream is still
clunking along at its original speed, and sending more streams will
not increase the speed of any one of them?

I'm inclined to think that the increase in speed is only affecting
the

additional stream(s) and not that original one.

Thanks.

George


To sign off this list, send email to listserv AT listserv.temple DOT edu and
type "signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with
this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or via RSS at
http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER




To sign off this list, send email to listserv AT listserv.temple DOT edu and
type "signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

To sign off this list, send email to listserv AT listserv.temple DOT edu and type 
"signoff networker" in the body of the email. Please write to networker-request 
AT listserv.temple DOT edu if you have any problems with this list. You can access the 
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER



--
George Sinclair
Voice: (301) 713-3284 x210
- The preceding message is personal and does not reflect any official or unofficial position of the United States Department of Commerce -
- Any opinions expressed in this message are NOT those of the US Govt. -

To sign off this list, send email to listserv AT listserv.temple DOT edu and type 
"signoff networker" in the body of the email. Please write to networker-request 
AT listserv.temple DOT edu if you have any problems with this list. You can access the 
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

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