Architecture assistance?

[VEBC]

Hi all...

Just looking for some help really... I have the following legacy environment:

Site A >> 1Gbps link Site >> Site B >> 1Gbps link >> Site C

Site B is the main site and has several buildings.

We currently have a 5 node VMware cluster connected to a FC SAN in Building A of Site B, and a second FC SAN in Building B of Site B which is currently primarily being used for testing and some backups of the VMware cluster. Two tape drives are also located in Building B and are used for the remaining backups of the VMware cluster. The two buildings are linked via internal FC, and the backup server itself is also in Building B and using SAN Transport to backup. We get good speeds from this to tape (8000-12000 MB/min).

What we were originally thinking of doing to make this better was implement Veeam and in addition to the backups occurring in Site B Building B on the FC SAN/Tape we would also deploy a server with 12 x 4TB NL-SAS disks in Site C and push backups to this. This would then give us the Veeam recommended 3-2-1 backups with two different types of media and one off data copy site. (We do move the tapes but its not instantaneous etc.)

However we are now being asked about site level resiliency in general which we do not currently have as all server/storage equipment is hosted in Site B.

So... with this in mind... we were wondering about moving one of the SAN's to Site C and running some new VMware hosts at this location. These VMware hosts would run VM's which have replication/redundancy built in at the application level (AD, Exchange DAG, DFS etc.).

At this point we could also utilize Veeam replication for some of the more mission critical VM's that are single instances such as application servers.

What I am wondering is, is this a good idea from the backup perspective? And how this would affect the backup process? As I am now limiting the throughput to the 1Gbps WAN connection if I backup to the SAN in Site C? How much of a big deal is this - I come from a world of weekly fulls and daily differentials or incrementals, and I know Veeam has better techniques than this that might mitigate my thoughts.

I do not want to overcomplicate things but I want to try and get the best possible setup in terms of resiliency and performance...

I know a lot of this varies on the amount of storage required, number of VM's etc...

So to summarise...

3 Sites connected through a central site with 1Gbps connections
2 SAN's
2 Tape drive units
Option to re-use/populate an existing 12 bay server with 4TB NL-SAS as another target

Thoughts?

[PTide]

Hi,

If you can move your second SAN to site C and place there a couple of hosts then you are good to deploy a replication of your mission critical VMs from your main site (B) to your recovery site (C). 1Gb link is ok for replication, especially if you use one of our advanced replication techiques. I'd also place your spare 12x4 server to building B at site B in order to have a backup storage that is not distant from your primary site. Also that would give you an option to take replicas from backup therefore reducing the impact of a snapshot on production.

Thank you.

[VEBC]

Hi,

I am with you about being able to deploy replication at this point, however what I am concerned about is the backup performance aspect?

Are saying I should backup to the 12x4 server located in Site B as the primary target?

I was originally considering backing up/Replicating to the SAN on Site C?

[skrause]

In most cases, you want your primary backup target location to be as close as possible to your production storage for the best performance.

You can then run a backup copy job that will take your backup files from your primary location to your fail-over location at a set interval. That way, the backup window on your production systems is not affected by the speed of your WAN link.

You can still use the SAN in Site C for the location to store the backup copies.

[VEBC]

Hi,

OK I think I am beginning to understand this architecture now.

So I am guessing idea for me is - Production SAN in Site B, Backup Repository in Site B (Using 12x4 server) which stores a small amount of recent backups (7-14 days retention etc.).

(Could I just use the Production SAN in Site B for this purpose - I know its not ideal but I could split the production and backup data disks into separate pools at the very least)

(Is there any calculator to estimate storage space required for backup repository? As I am limited by the server for 12 disks...)

I then use Backup Copy function to push this data for longer term storage onto SAN in Site C?

From where do I then use Remote Replica from Backup option? Site B Backup Repository, or Site C Backup Copy?

[PTide]

So I am guessing idea for me is - Production SAN in Site B, Backup Repository in Site B (Using 12x4 server) which stores a small amount of recent backups (7-14 days retention etc.).

Sounds good. If you can connect any Windows physical server to your production SAN via iSCSI/FC then you'll able to utilize direct SAN approach allowing to perform backups really fast.

Could I just use the Production SAN in Site B for this purpose

Doesn't sound that good. Why would you want to do that? It will cause a high load to your storage and might result in an overall slowdown of business processes.

Is there any calculator to estimate storage space required for backup repository?

Please check this

I then use Backup Copy function to push this data for longer term storage onto SAN in Site C?

Correct.

From where do I then use Remote Replica from Backup option?

That depends. If you copy backups every time a new backup appears then it would be better to use backup copies - the same restore points but much closer than your original backups. In case you are going to copy your backups less often that they are created then I'd recommend to replicate form original backups in order to have the most recent replica state.

[albertwt]

Does the Backup copy from the Original Backup file is supposedly to run faster than running it from the actual Storage Array Snapshot VM backup ?

[foggy]

You cannot compare, different components are involved in the process.