What´s your infrastructure for Veeam Backup?

[aalmeidadc]

What´s your infrastructure (Physical and Logical) to support your backup?

This is my:

-7 Physical Servers IBM X3650, 2 procs X5680 6 cores, 96 GB each

-Production Storage: IBM XIV

-Backup Storage: IBM V7000 (Linear SAS Disks)

-Communication: SAN 4GB

- Production Data: 17 TB (Full Size VM´s)

-1 Veeam Backup Server 6.5 (is a Backup Proxy too). 6 cores X5680, 12GB Memory, 2 backup repositories

-1 Veeam Backup Proxy. 6 cores x5680, 12GB memory, 1 backup repository

-12 Backup Jobs separated by type of application (such as Application Servers, Infrastructure Server (AD, DNS, Radius…) , SQL Database Servers, Oracle Database Server and so on…) with a maximum of 2 concurrent job by Backup Repository and

-3 concurrent job by Backup Proxy Server)

Troughput (Incremental Backup, with compress and dedupl): Almost 1 Gb/s

Real (Disk troughput): Around 130Mb/s

Configured Options:
- Type: Incremental (Synthetic Full Weekly)
- Inline data deduplication: Activate
- Compress Level: Optimal

- Backup Data: 8,5TB ( 8 days of retention)

[dellock6]

Hi Andrè, really nice design, how many VMs are in there inside those 17 Tb??? I really envy your V7000 as backup storage, read my postand you will find out why...

My setup has a different design, sharing here some info about it:

- around 1000 VMs and 56 TB
- veeam backup on a dedicated VM with only console, enterprise admin
- SQL server 2008 on a dedicated VM (also holding vCenter and other management databases)
- iscsi storage all around, via LeftHand both physical and virtual
- 7 proxies, all virtual, each with 4 vCPU and 4 Gb ram, windows 2008 R2 core edition
- reverse incremental, 7 days retention, no syntethic nor full active
- job divided by resource pool (no choice here, vCloud creates a RP for each tenant, so every backup set is dedicated to each customer)
- (this is by far the bottleneck I'd like to upgrade soon) 4 repositories made with physical servers, each receiving 2 jobs, 8 cores and 12 Gb ram, using a mix of local storage and a couple of HP P2000 in local attach mode
- Processing speed: 332 MBs, almost 8 hour of execution

I'm looking around for really faster backup storage, I had some ideas about creating new "VeeamPods" to act as repository, but what I do not really like is the backup jobs beeing stored locally on the repository server; you loose a server, you also loose the repository. Your design with a dedicated external backup storage sounds really good to me, you can also swap physical repositories keeping the same backup storage, and even mount a given lun holding backups to another repository at will. Nice indeed.

Luca.

[ddayton]

I'm on a real low budget. I am running ESXi 5.0 (free) on a Supermicro Server. The server has dual 6-core 3.47GHz Xeons, 96GB RAM, 2 on board 1G NICs and 2 dual port 1G PCIe add-ons. The server has a built-in 8-port SAS/SATA with RAID contoller and we are running with four 500GB SATA and four 2TB SAS drives. We have 10 VM's running for production.

1 - Linux CENTOS 6.3 (64bit)
3 - Windows 7 Enterprise (32bit)
1 - Windows Server 2003 (32bit)
2 - Windows Server 2008 (32bit)
3 - Windows Server 2008 (64bit)

I purchased the Veeam Enterprise Bundle with B&R and One. I installed two 4TB NAS Servers using FreeNAS setup as iSCSI to the VM that is running Veeam B&R and One. One NAS is in our datacenter and one in our DR site. I am currently doing weekly full to both and incremental to DR and reversed incremental to the datacenter. I created 10 backup jobs half targeted to each NAS split as listed by OS hoping to get the most for dedup. I plan to also add out LTO3 Tape Library to the NAS Server for targeting longterm storage.

I have asked for a budget next year to add a second VMware server in the DR and at that time get the paid license versions giving me vMotion and support.

[Yoann Castillo]

I am a architect in a SSII and I installed recently at a client, Veeam 6.5.
- ~ 100VMs
- ~ 15TB data
- Storage DS3500 backup+prod (FiberChannel)
- 2 sites: Prod + Backup

The production site sending backups to the backup site.
Veeam servers:
- 1 server on the backup site
This server is a VM and is connected "DirectPath I / O" on FC card via the infra FC storage.
Proxys servers are installed on the production site to handle the load if necessary.

Backups are therefore DirectAccess (via FC) ... The debit is around 3-5Gb / s ... )


It's true that the DirectPatch I/O is contrégnante because the vMotion or HA is not allowed ... But no matter, because Veeam proxies are configured to operate in a degraded mode if needed.

This is an interesting config, I think.

cdt,

[agtcovert]

Production ESX Servers:
3 x Supermicro somethings, 2xE5645 @ 2.4
2 x Dell R710 2xX5650 @ 2.8

~160 VMs
~15TB

-Backup storage: 24 bay Supermicro server/ Windows 2008 R2 / plus SQL Server 2008 R2 for Veeam database (plus other infrastructure related databases).
Server is the master Veeam server / backup proxy
Storage is 2xRAID10 mirrors (around 16TB useable total, used R10 to avoid long rebuild times)
1 additional Windows 2008 R2 proxy server, 4 threads

Production Storage: EMC VNX 5300 & VNXe 3100
1GB ISCSI

Backup processing runs around 80-90MB/sec
Some servers are backed up nightly only
Other customer VMs are backed up (reverse incremental) every 4-6 hours due to high change rate/recoverability requirements.
Synthetic fulls on the weekend split between late Friday/Saturday

The setup works very well for us. Love the product, and it's saved us a couple of times and made customers happy when we can rewind in time quickly if they make a mistake, etc.