Comprehensive data protection for all workloads
Post Reply
Dejavoodoo
Lurker
Posts: 2
Liked: never
Joined: Jan 16, 2014 11:17 am
Full Name: DejaVoodoo
Contact:

Hardware Help

Post by Dejavoodoo »

Hi guys,

We've being using Veeam 7 for a while now and an opportunity has arisen to move the server onto its own dedicated piece of tin rather than a VM. Now, I've had a few issues with WAN Accelerators and iSCSI so I'm thinking about the following and wondered what you experts feel:

So first, a bit about our infrastructure:

76 vSphere ESXi 4.1 hosts, running 3 VMs (RODC, FP & TS) per host connecting to WAN on standard ADSL+ (we will be migrating to Hyper-V this year :( )
All VMs are Windows 2008R2 and deployed from the same image
Veeam Proxy + Accelerator installed on RODC (no vPowerNFS)
Reverse Incremental job to backup local FP to NAS device (average server size 450Gb, daily changes to data ~300Mb)
Backup Copy job to copy above to central SAN at HQ

Now, in the past, I used a QNAP 12Tb iSCSI NAS to host the Target WAN Accelerator and 9 times out of 10 it s**t itself (RPC errors, corrupted cache etc) and caused lots of failed jobs so I ended up doing a direct copy which has been working fine, its just not quick enough.

Moving forward, I have a HP Proliant DL360p Gen8 with 8x1Tb 7.2k SAS disks installed in RAID5, Xeon 6 core E5-2360 @ 2.3GHz and 32Gb RAM. I plan to install Windows 2012 R2 onto this box and use every inch of available local disk storage as my WAN Accelerator Cache. Backups will go to a 21Tb Equallogic PS4000E via a dedicated qLogic HBA with dual Gb links

Can anyone see me having any issues with that?

Thanks in advance

DV
foggy
Veeam Software
Posts: 21069
Liked: 2115 times
Joined: Jul 11, 2011 10:22 am
Full Name: Alexander Fogelson
Contact:

Re: Hardware Help

Post by foggy »

Dejavoodoo wrote:I plan to install Windows 2012 R2 onto this box and use every inch of available local disk storage as my WAN Accelerator Cache.
Just wanted to note that WAN cache size neither does noticeably impact WAN Accelerator performance nor WAN bandwidth savings (the impact of the larger cache will not be significant). What provides better effect is separating global cache from repository by placing them on different disks (what you're already going to do) and using fast SSD disks with a low latency profile for WAN cache.

Also, note that while you can use the same target WAN accelerator for multiple source sites, each pair of accelerators will use its own cache on the target side.
Gostev
Chief Product Officer
Posts: 31455
Liked: 6646 times
Joined: Jan 01, 2006 1:01 am
Location: Baar, Switzerland
Contact:

Re: Hardware Help

Post by Gostev »

The above is correct as it relates to WAN Accelerator data processing performance ONLY.

Larger cache does of course improve WAN bandwidth savings, but there are a couple of caveats to be aware of:
1. Diminishing returns. The more disk space you add to the global cache, the smaller the improvement you will get (on top of what you already have).
2. The large cache you create, the faster storage you need to use for the cache, or it will impact data processing performance.

I always recommend using SSD for the global cache regardless of its size, but if you are looking at having much larger cache than the default 100GB, the you most definitely want to place it on SSD. Not really a problem considering how cheap they are these days (comparing to bandwidth).
Dejavoodoo
Lurker
Posts: 2
Liked: never
Joined: Jan 16, 2014 11:17 am
Full Name: DejaVoodoo
Contact:

Re: Hardware Help

Post by Dejavoodoo »

Thanks for the advice chaps.

SSD's on order....
Post Reply

Who is online

Users browsing this forum: Bing [Bot], Google [Bot], jorgedlcruz, mbrzezinski and 178 guests