Comprehensive data protection for all workloads
Post Reply
bertrand82
Lurker
Posts: 2
Liked: never
Joined: May 10, 2010 10:55 pm
Full Name: Chris Ebert
Contact:

esx -esx local seed

Post by bertrand82 » Mar 16, 2011 10:30 pm

Hi All,

I have some questions on the below scenario:

Scenario

We have two ESX 4.1 hosts (Host A & Host B) connected to a Fibre SAN and both have access to the same LUNS. Veeam 5 is installed in a Win 2003 VM running on Host B. We are try to achieve a local replication of VM's from Host A to Host B (local storage) via FC using the direct SAN method. We want to use the initial local replication as the "seed" replication with a secondary replication occurring over the WAN to an offsite location.

Questions

1. Given the scenario above, will the direct SAN method work with Veeam installed on a VM or do we need to install Windows and have it directly connected to the SAN. I assume in this scenario we are using the virtual appliance method - is this correct?
2. Is the scenario above the best method to achieve this?
3. Does Veeam have any automated support for this or would it require two Veeam replication jobs and effectively manage it manually.(i.e. the primary job locally creating a secondary job to the remote data centre which essentially is part of the same job rather than us running two replication jobs (vm > local & local > remote)

Thanks in advance

Chris

Gostev
SVP, Product Management
Posts: 24431
Liked: 3406 times
Joined: Jan 01, 2006 1:01 am
Location: Baar, Switzerland
Contact:

Re: esx -esx local seed

Post by Gostev » Mar 16, 2011 10:46 pm

Hi Chris,

1. Direct SAN method cannot be used with Veeam installed on a VM in case of FC SAN. Virtual appliance mode is the way to go.
2. One gotcha is that you will not be able to use leverage changed block tracking when replicating a replica. However, since replication is usually WAN link speed bound, and local storage is not loaded by any tasks, this may not be a problem. Otherwise, you could just setup 2 jobs with the same source but different destination. This is actually more reliable, because you get the data straight from source, instead of replica.
3. You need two jobs.

Thanks!

Post Reply

Who is online

Users browsing this forum: Bing [Bot], falkob, ferrus, FSenturk, Google [Bot], restore-helper and 44 guests