cancel
Showing results for 
Search instead for 
Did you mean: 

Wow,upgraded from v.10 to v.10d and now backups are slow and use more tapes

Natalie_Caciopp
Level 3
Seven days ago, we upgraded from Backup Exec version 10 (5520) to Backup Exec version 10d (10.1 5629). I upgraded all my Remote Agents to 10.1, I upgraded my tape drives to driver 5.1.1.1 (driver date 1/31/2006).

Since the upgrade, my throughput rates have dropped significantly. For example, I went from backing up our SQL servers at 2.8 Gb/min down to 1.6 Gb/min. Exchange went from 2.5 Gb/min down to 1.8 Gb/min. This is a big difference in the amount of time it takes to do a backup.

I not only have the speed degregation issue, I also have a problem with the backups using five extra LTO2 tapes for this past weekend's full backups. There was no significant data increase to cause Backup Exec to use even one more tape but all of a sudden, it requires five extra tapes for our full jobs. I have the same compression selection as before so I am guessing that ver 10d (or the tape drive drivers) doesn't handle the compression as well and that is why it is using the extra tapes...

So with Backup Exec 10d, our backups are running much slower and it uses many more tapes for the same amount of data. I have read past posts and have talked to a couple other contacts who have the same exact issue I have. I really do not want to roll back to version 10, that will be a last resort but I will do that if nothing else fixes the problem.

I would love to get some feedback on this, maybe someone had my same issue and knows of a simple resolution...I hope. :)

Thanks in advance to all...
14 REPLIES 14

padmaja_rajopad
Level 6
Hi,

Try running the job to a B2d folder and compare the results.


If you are using a standlaone drive, you may try using NTBACKUP for comparison purposes..




NOTE : If we do not receive your reply within two business days, this post would be marked �assumed answered� and would be moved to the �answered questions� pool.

Ken_Putnam
Level 6
You are not alone. And as you can see from the "official" response, Veritas/Symantec still refuses to acknowlege that this problem which has been reported amany,many times, even exists.

One thing that you can check. From the Devices tab, right click the tape drive/Properties

What are the settings for block size, buffer size, number of buffers, and write through?Message was edited by:
Ken Putnam

Typos

SirEbrum
Level 3
Partner
I'd be interested in finding out if the tape device driver was updated.

Certainly an option would be to attempt using the OEM's Tape device driver (HP, IBM, etc), run Tapeinst.exe, leaving the manufacturers driver in place, and see if speed/capacity increases.

Also, HP has a little utility called "tapeperf", which can be used to test drive performance, regardless of the make model. As I understand, it generates compressable data at 2:1 and 3:1 ratios in memory and saturates the SCSI bus. Differences in performace between the two drivers (Veritas vs. OEM) would be interesting!

Good luck Natalie

- Tom

Sharvari_Deshmu
Level 6
Hello,


We request you to update us regarding this issue.


Thanks,

NOTE : If we do not receive your reply within two business days, this post would be marked assumed answered and would be moved to answered questions pool.

Ken_Putnam
Level 6
One thing I forgot to ask -

From the devices tab, right click the drives\Properties

What are the settings for

Block Size
Buffer Size
Number of Buffers,
Single Block Read/Write
SCSI Passthrough

Several other people that have upgraded find that Block size has been set to 1K!!!!

Natalie_Caciopp
Level 3
I have several jobs running and they are a combination of backup-to-disk and straight to tape jobs (using the same servers). The slowdown is across the board, all jobs are affected the same.

Here are the tape drive properties:

block size: 64K
buffer size: 64K
buffer count: 10
high water count: 03:28 PM 3/27/2006
read single block mode...unchecked
write single block mode...checked
read scsi pass-through mode...unchecked
write scsi pass-through mode...checked

I actually called Symantec on the tape drive properties a couple weeks ago and they said to not change the default tape drive settings.

This past weekend, I set several jobs to use different compression settings and that actually made things worse so I will be setting those back to what I had before. My full jobs this weekend used five extra tapes again. That is ten extra tapes in two weeks...that equals $500 extra dollars to run version 10d in two weeks. :(

I need to call Symantec but I know it will be a long troubleshooting process and I need to find time in my schedule for it.

Do you know, if I do end up going back to version 10.0, when I uninstall the RA from all the servers, do I have to reboot the servers after that uninstall before I can install the version 10 back on?

Ken_Putnam
Level 6
I'd de-select single block write
I'd bump the buffer to 128K
And I'd deselect both pass-thru settings


See http://seer.support.veritas.com/docs/259030.htm for a little more detail

If that improves things, you could try a buffer of 256K, but that probably won't make that much difference.

John_Richards_2
Level 3
For what it's worth, and it's probably of no help, but just in case, I switched over from 10.0 to 10d and my BTD and tape duplication jobs are identical in times, give or take a few seconds on a multi-hour job. I'm using DLT tapes in a Quantum Superloader instead of LTO tapes. I hope you find the cure. That is sure an expensive upgrade for you.Message was edited by:
John Richards

mark_green
Level 3
Hi i have had exactly the same problem .
my backups are slowing and using double the amount of tapes.
my full weekend backup used 30 tapes instead of 19.
I am going back to version 10.0,, and uninstalling 10.d.

Support could give me know answer to this problem, saying it was my hardware.

no its not hardware is fine.

Natalie_Caciopp
Level 3
Well, I ended up reverting back to Backup Exec 10.0. The throughput rates are back to what they used to be.

I will find out this weekend if my backups stop using the five extra tapes like it did with 10d. I will definitely report back on this board next week with a follow up on that.

Ken, after my weekend backups run, I will toy around with changing the settings on my tape drives to what you are suggesting. If I can gain some extra speed that way, I am all for it.

I will wait until a hotfix or the next version is released to upgrade again. I did call Symantec when I uninstalled and reinstalled to the old version because I had some issues and I did notify them of the reason why I was doing this. Hopefully there will be enough support calls and complaints about this problem so that a fix can be issued soon.

Natalie

Natalie_Caciopp
Level 3
After reverting back to 10.0, my backups this weekend did NOT use the five extra tapes that 10d required. Strange, I am using the same compression settings and it is the same data, I don't know why 10d used more tapes like that.

Systems_Admin
Level 3
We upgraded to 10d the week it was available and it was the most seamless upgrade yet. No decrease in speeds, in fact if anything there was an increase to our DB/Exchange backups. I'm a tad annoyed at the continued security vulns in all the remote agents and the required reboots of all my servers, but what can I do? BE has saved my *ss numerous times. 10d is where the product should have been 5 years ago.

Paul_Scalzo
Level 3
Can you give specifics? how fast was your 8.5 system compared to this version? Exchange speeds etc...

shweta_rege
Level 6
Hello Natalie,



Could you please Update us on the issue?



NOTE : If we do not receive your reply within two business days, this post would be marked assumed answered
and would be moved to answered questions pool.

*************************************************
************************************************