Veeam Backup on Hp Microserver

[lombardi.aless]

Hi,
anyone had already use for a very small project (1 host with esxi or HV 2012 and 2-3 vm), one HP Microserver as a backup server with Veeam installed?
I know the performance of this server is very poor but is also very cheap and for a environment like, this where you cannot make veeam in a VM, i think that you can reach your purpose with less.
Any suggestion or opinion?
Thanks
Alex

[Gostev]

Hi Alessandro, what are the specs for this server? Generally speaking, you definitely should not be worried about the backup server performance when talking about backing up just 2-3 VMs. Thanks!

[lombardi.aless]

This is the specifications:

http://h10010.www1.hp.com/wwpc/us/en/sm ... html?dnr=1

it's a small form factor server with an AMD Turion II processor that can reach 16 gb of ram and hd sata.

[Vitaliy S.]

Hi Alessandro,

16 GB of RAM will not be required, 4 GB should be more than enough. As to the CPU horsepower, then I guess it should also be fine. However, if you start missing your backup window and your CPU usage will be maxing out while running backup jobs, please consider reducing the compression level to have better job performance rates.

Thanks!

[dellock6]

Hi Anton,
I'm not totally convinced about the memory requirements for Veeam, especially the Repository role. I've done extensive tests in our datacenter, and running Reverse Incremental jobs against VBK files bigger than 500 GB each is consuming a quite amount of memory.
It's not the case of Alessandro and other really small design, so you are right here, but it's not an absolute truth. I've seen the Disk Cache service on Windows 2008 R2 eating more than 1.5 Gb RAM to open and manage that VBK file, so if you think about a repository receiving multiple backups at once, memory suddenly becomes an issue.
I've wrote a blog post about it, also with some tweaks using RamMap and CacheSet tools, it's going out in few days. (WARNING: run these tools at your own risk!). But at the end, the real solution is to give the repository service more ram...

Luca.

[Gostev]

Hi Luca - even though it was not me who posted this I fully agree with the statement considering that what is being discussed in this topic is a very specific, tiny deployment. Vitaly's response was not meant to provide generic recommendations for shop of any size, but to answer specific question.

For large deployments, you are right. I know customers who are using servers with 64GB for backup repositories (due to multiple concurrent jobs hitting those)...

[dellock6]

Ops, sorry, I rushed through the thread too much...
Yeah, we are both right, only wanted to clarify for future readers, don't want to listen them reading here, undersizing repositories and then come back complaining.

Btw, today at a customer we started a new Veeam physical backup server, FC directSAN mode at 8 GBps, he gave me what he called a "spare" server, was a Dell R710 with dual X5000 quad core Xeon, and 128 Gb of RAM. Pretty insane!!!

[Gostev]

I'd like to have such a "spare" haha!

[poulpreben]

I have an N36L with 4x 2 TB SATA drives and 8 GB of memory for my home NAS. It is, however, running FreeBSD with ZFS (raidz) underneath. It can easily sustain a GbE connection via CIFS, NFS and AFP with that amount of memory.

I have tried running ESXi on it, but it is very slow. The main problem is the lack of a real RAID controller to span the VMFS across all four drives. This could be solved by adding an additional P4xx controller from HP, which I have had good results with.

Anything you'd like me to test?