cancel
Showing results for 
Search instead for 
Did you mean: 

Hardware requirement Enterprise Vault 8.0.3

Contonso
Level 4
Hi there,

We are running Enterprise Vault now nearly for one year. In the last 2-3 months we are fighting with backup issues at our constallation.
I would like to give you a few details about system and users and would like your point of view about this.


- Enterprise Vault for Exchange 8.0.3
- We have about 2200 mailboxes
- Vault Cache is enabled for aprox 2000Users (15% free disk space)
- The average size of the mailbox per user is about 500 MB
- We did not do a PST Migration
- Archive all items older than 30 days.
- collections enabled (20 MB) collect items older than 5 days
- 600GB Vault Store in 5 Partitions


Deatails about server hardware:
- HP Proliant DL 18 G6
- Intel XEON E55202.27GHz  Processor
- 4 GB RAM
- 6 X 750GB SATA Local HDDs (RAID 6)


Archiving runs from mon-fri from 4pm-10:59pm
and sunday from 8:00pm - 5:00am

Incr. backup starts mon-thur at 11:00pm (open end) 
Full backup starts fri 11:00pm (open end)

Backup client is  Legato Networker.


As I said we are always running into Backup issues. The Full backup takes more than 1200minutes (sometimes longer) and runs sometimes into an error.
The Incremental backup runs sometemies for 3-5hours and stops with and RPC connection error.

I think that the system constellation especially with local SATA HDDs are not chosen well, because if there are multiple read/write accesses, the system is very slow.
Connection with RPD is nearly impossible

My question is, what is your opinion on this. Does it make sense to use such an hardware especially local HDDs at more than 2000 users?
What is your recomendation?  What is your exerince.

Any help would be appreciated.

Thanks







3 REPLIES 3

Liam_Finn1
Level 6
Employee Accredited Certified
Local 15K disk are very fast if you have enough disk to meet the IO demands of the application

With a RAID 6 you actually only read and write from 4 disk so at 15K disk your IOPS is only 720 for all transactions which pretty low.

Depending on the number of recalls and the size my guess is that the disk is definitely a bottleneck that you need to consider

Also remember that from what you are saying your MSMQ and DATA and INDEXES are all on the same disk which means those disk are working very hard indeed to meet the request for data. Also your page file is either on the same disk or on your C drive which is not optimal.

I think you should consider some overhaul on the hardware side to meet performance.

Chai
Level 4
Employee
Agreed with previous comments.. Definitely run Performance counters against CPU, MEMORY, DISK to see what is causing this slowness.. Especially when RDP is having issue

Contonso
Level 4

Thanks for your responses...
 
Liam, you wrote:

Also remember that from what you are saying your MSMQ and DATA and INDEXES are all on the same disk which means those disk are working very hard indeed to meet the request for data. Also your page file is either on the same disk or on your C drive which is not optimal. 


 Pagefile is at drive C:\. Two days ago I got this message in the event log.


The system is running out of Virtual Memory - free memory by closing any unwanted applications

Enterprise Vault will be shut down when 95% of Virtual Memory is used


The PF usage was at 6,4 GB and drive C:\ (20 GB) was nearly full. This above message appeared during the A1 que was handled.

Chai, you wrote:

               Definitely run Performance counters against CPU, MEMORY, DISK to see what is causing this slowness.. Especially when RDP is having issue

Can you tell me at what counters exactly to run the performance test? In "perfmon" I have lot of counters.

What I would like also to know is:

What would be the hardware suggestion for our requirements with more then 2000mailboxes and our archiving strategy and VC build etc.

Centera, Netapp or SAN? Or local system with local HDDs?


Thanks in advance