UNSOLVED
TL2000 LTO4 and Backup Exec 11d
I am curious what kind of transfer rates I should be getting from our LTO4/TL2000 with Backup Exec 11d. Server is a PE 2950 Gigabit with Perc 6/E attached MD1000 6 drive Raid 5 10k 400gb SAS drives. TL2000 is attached via SAS 5/E.
This is all brand new equipment.
I do not find it to be much faster than our LTO2 Exabyte with BE 10d.
11d seems to run at 25% of the rate 10d would run at for remote backups.
This is all brand new equipment.
I do not find it to be much faster than our LTO2 Exabyte with BE 10d.
11d seems to run at 25% of the rate 10d would run at for remote backups.
DELL-Bob D
899 Posts
702
0
Posted March 5th, 2008 21:00
With that setup I would expect to see a minimum of 70MB/sec non-compressed data.
What sort of throughput do you see?
Boblocal backup 1800 - 2200 MB/Min
remote backup 500 - 600 MB/Min
Both those numbers are with hardware compression. Dell ran there Arm test on the drive and came up with 70 MB per sec but that I believe was copying 4 gigs of the same character which is not a very good test.
Spent 3 hours on the line with Symantec (this was a joke) They could not get 11d to work with anything but the Symantec tape driver and put the blame on IBM.
I am about ready to send the TL2000 back and get something that works well.
I am using the latest Symantec driver for the tape drive. Symantec was only trying to use the IBM driver to see if it was any faster.
Are you able to confirm if the Firmware/drivers are the latest for the list components?
SAS 5/e Driver: 1.26.5.0
Firmware: 00.10.51.00.06.12.05.00
PERC 6/e Driver: 2.14.00.32
Firmware: 6.0.1-0080
I will test backup to disk and to tape today without compression.
DELL-Bob D
899 Posts
702
0
Posted March 6th, 2008 15:00
Does the throuput # for the local backup increase when you use the Symantec driver? You should not be using IBM drivers for the drive if you are using BEWS software.
If you haven't already try running a LOCAL backup using NO COMPRESSION and the symantec tape drivers. Be sure the backup is atleast a 10minute job to insure the drives have ample time to get to top speed and sustain that rate for a bit. The reason I stress NO COMPRESSION is that the compression is dependant on the compressability of the files..since you do not know the true compresability of your data set you will never be able to determine what a true speed should be thus never being able to pin point the bottleneck. Using NO COMPRESSION insures you that the top speed of the drive is 120MB/sec.. take 10-15% off that # for OS overhead and you now have a target speed that should be achievable with any data set.
If your #'s are still sitting in the 36MB/Sec range you need to determine if the data coming from the source is capable of delivering at this rate... from your previous post to me it sounds like you have plenty of horsepower but you may want to create a backup to disk job and eliminate the tape drive and the 5/e controller from the picture. Run the same job as above but send the data from the disk array to a backup disk folder that is local to the server. The drives should have absolutely no problem hitting above 36MB/sec. This will give you the "from disk array to server" rate. If this # is low I would start looking at drivers for the MD and the controller it is attached to.
Before you start any of this be sure you have the latest drivers/firmware installed for the 5e card.
Be sure you are using LTO4 media.
I have run this configuration numerous times and can assure you 100+MB/sec is achievable with the setup you have.
The remote backup is dependant on too many factors. When you figure out your local problem you can begin to tackle the remote issue. For that I would begin by getting a system plugged into an isolated switch with the server... network traffic will always kill good performance and in my experience is usually the problem.
Not how you want to begin with new h/w I understand but I believe you can get this running as it should and you will be pleased with the TL product.
DELL-Bob D
899 Posts
702
0
Posted March 6th, 2008 16:00
Interesting...
The PERC card is up to date w/ both drivers/firmware.
The 5/e however has a "newer" rev than what has been posted on support.dell.com as of 12/07... not sure what that rev is from I will need to do some asking about that one. I would go ahead and run the one that is posted 1.24.4.0. The pckg # is R170398.... hopefully you can pull it from this link : http://support.dell.com/support/downloads/format.aspx?c=us&cs=19&l=en&s=dhs&deviceid=9840&libid=45&releaseid=R170398&vercnt=6&formatcnt=0&SystemID=PWE_2950&servicetag=&os=WNET&osl=en&catid=-1&impid=-1
The f/w on the 5e card is good.
I will try to track down this 1.26.5.0 driver and run a quick backup with it to see the perf #'s I see.
Bob
DELL-Bob D
899 Posts
702
0
Posted March 6th, 2008 17:00
The best I can tell the 1.26.5.0 is a Windows Longhorn driver (windows 2008).
Is that the driver version that was loaded with the server when you got it?
Bob
The link I sent you was for the x32 version of the driver here is the x64 version: http://support.dell.com/support/downloads/download.aspx?c=us&cs=19&l=en&s=dhs&releaseid=R170404&SystemID=PWE_2950&servicetag=&os=WX64&osl=en&deviceid=9840&devlib=0&typecnt=0&vercnt=6&catid=-1&impid=-1&formatcnt=3&libid=45&fileid=232118
Message Edited by Dell-Bob D on 03-06-2008 02:00 PMDELL-Bob D
899 Posts
705
0
Posted March 6th, 2008 18:00
I am getting ready to run the driver (1.26..) now...
Just to be clear... this is Windows 2003 correct?
If this is the driver that came factory loaded on your server I need to find out how it is getting there... can you please confirm where the driver came from?
cybervex
11 Posts
702
0
Posted March 3rd, 2008 18:00
I guess this forum is useless!!!!!