cancel
Showing results for 
Search instead for 
Did you mean: 

Overhead for Enterprise Vault

WayneCierkowski
Level 3
We have just installed EV and have been testing it. We have noticed that we see an overhead of around 40-50 seconds for a file to come back from the vault. Just wondering if this is the norm for the product. I think my management was under the impression that a archived file, even a small file, would only take a "Couple of Seconds" to come back from the vault. 

Thanks in Advance...
NetBackup 8.1 at 10 sites
IBM 3584 LTO 6 - MSL 6480 LTO 6 - MSL 4048 LTO 6 - D2D
7 REPLIES 7

Liam_Finn1
Level 6
Employee Accredited Certified
It is not the norm. Performance issues can stem from many items such as lack of memory, underpowered SQL server, Not enough spindles to service the Disk IO requests

Have you tried running performance monitor to see where you are experiencing bottlenecks

TonySterling
Moderator
Moderator
Partner    VIP    Accredited Certified

I know you just installed but could you clarify what version of EV, what version of Windows Server and SQL?

Also, did you enable Collect and Migrate for your partitions?

Regards,
Tony

Rob_Wilcox1
Level 6
Partner
I assume from your description that you're doing File System Archiving, rather than Exchange Mailbox Archiving?  What sort of size were these files originally, or is it all files that you're experiencing the delay on?
Working for cloudficient.com

WayneCierkowski
Level 3
Thanks guys for the response's.. Now let me answer some of the questions...

I haven't run any performance monitors as of yet. The EV server also holds the SQL server. It is a blade unit with two Intel Xeon Quad processors, 2.5 GHz for each core.The box also has 12 Gig of main memory. Disk layout is as follows, C drive OS and installed EV software, E is for MSMQ, F holds the SQL databases, G holds the SQL Logs, H is the single vault store I have defined, I holds the Index for EV and S is where SQL is installed too. One other thig concerning the disk's on the system. The F, G, I and H all reside on a single volume that is on our SAN, HP EVA. Fiber Connection with 2 GBs fiber cards. Drives C, E and S are contained on the local drive of the blade.

Yes I am doing pure FSA type archiving. In out testing we only archived around 152 files right now. And they are small. Most less the 500 MB's, if not smaller. 

Version of EV is 8.0 SP4... OS is Server 2008 with SP2 applied. SQL is SQL Server 2008.

I really appreciate all the help I can get with this.

Thanks Again,

Wayne
NetBackup 8.1 at 10 sites
IBM 3584 LTO 6 - MSL 6480 LTO 6 - MSL 4048 LTO 6 - D2D

ZeRoC00L
Level 6
Partner Accredited
Just to be sure, are you running Windows 2008 64 bits or not ?

WayneCierkowski
Level 3
Yes 64 bit.... 
NetBackup 8.1 at 10 sites
IBM 3584 LTO 6 - MSL 6480 LTO 6 - MSL 4048 LTO 6 - D2D

Liam_Finn1
Level 6
Employee Accredited Certified
Let me guess, you are sharing the EVA with everyone else who sites in the same building as you and it is used for File, CAN and others

Run perf mon and watch the Disk transfers as well as disk queue length

My guess is that where your bottle neck is.

Here is my reasoning for this.

We used to have an EVA and because of the way the EVA does it VLUN's you are competing for IO on the disk. The slowness is doe to everyone else including any backups you run eating your disk IO and it is killing your EV environment.

I speak from experience. HP EVA is garbage as a SAN device for any high IO requirement unless it is 100% dedicated to your application.

An EVA when you create a VLUN creates a stripe across all the disks in the EVA. This in theory gives you lots of IO and it does as long as no other VLUN's exist and there is no IO competition on the disk. In a shared EVA the IO is not dedicated and users file access eats the IO and slows EV to a crawl