Discussions specific to the VMware vSphere hypervisor
Post Reply
ted
Novice
Posts: 6
Liked: never
Joined: Mar 22, 2011 4:07 pm
Contact:

Veeam Server Location

Post by ted »

Hi,

I'm thinking of implementing veeam and have some questions:

Two sites - Prod and DR
Prod site has 4 x ESXi 4.1 servers and 60 VMs at the moment the DR site has 3 x ESXi 4.1 servers that i want to replicate the VMs to. I am thinking of putting the veaam server in the DR site. Are there any potential issues with this, reason for having it here is if the prod site fails i can still implement failover. Any performance or other issues with having the veaam server in the DR site?

Thanks

Alexey D.

Re: Veeam Server Location

Post by Alexey D. »

Hello,

Here is a good topic which should cover your questions: Veeam Design Recommendations. Thanks!

Vitaliy S.
Product Manager
Posts: 24243
Liked: 1859 times
Joined: Mar 30, 2009 9:13 am
Full Name: Vitaliy Safarov
Contact:

Re: Veeam Server Location

Post by Vitaliy S. »

Hello Michael,

Just wanted to add that with ESXi target, Veeam backup server should be located on a target side, for sure. This is the most common practice among our customers. Such setup should give you better performance rates compared to a push replication model and ESXi hosts.

Thank you.

ted
Novice
Posts: 6
Liked: never
Joined: Mar 22, 2011 4:07 pm
Contact:

Re: Veeam Server Location

Post by ted »

Thanks for the quick response.

i see that one of the cons in this scenario is that i will loose network compression. Do you know what the compression ratios are? Would just like to know if this would make a big difference.

ted
Novice
Posts: 6
Liked: never
Joined: Mar 22, 2011 4:07 pm
Contact:

Re: Veeam Server Location

Post by ted »

Thanks Vitaliy. Looking for clarification on DR testing but i think this should work from reading other threads.

In a DR test we break the links between the sites then:
Run a failover test using veeam - assuming the vNetworks are named the same, i just script the IP address change of the VMs so that they are in the DR subnet
Test servers using clients to access applications
Finish test and "Undo Failover"
Bring back up the link between the sites.

Will replication re continue ( i didn't stop it just brought the site link down) or should i stop it before taking the link down and then restart it after the test?

Vitaliy S.
Product Manager
Posts: 24243
Liked: 1859 times
Joined: Mar 30, 2009 9:13 am
Full Name: Vitaliy Safarov
Contact:

Re: Veeam Server Location

Post by Vitaliy S. »

Michael,

Provided that you have followed the steps mentioned in our F.A.Q. then you should be able to continue your replication job.

As regards compression and deduplication ratio, then this only applies to rollbacks (VRB files), because an original, unchanged .vmdk should be presented to a target host. So if I were you I would definitely place the backup server on the target site.

Thank you!

ckd5150
Influencer
Posts: 10
Liked: never
Joined: Dec 05, 2009 7:27 am
Full Name: Carl
Contact:

Re: Veeam Server Location

Post by ckd5150 »

Vitaliy S. wrote:Hello Michael,

Just wanted to add that with ESXi target, Veeam backup server should be located on a target side, for sure. This is the most common practice among our customers. Such setup should give you better performance rates compared to a push replication model and ESXi hosts.

Thank you.

Why is it faster for a pull instead of a push?

Gostev
SVP, Product Management
Posts: 26679
Liked: 4268 times
Joined: Jan 01, 2006 1:01 am
Location: Baar, Switzerland
Contact:

Re: Veeam Server Location

Post by Gostev »

Because chatting with target ESXi happens over LAN this way.

ckd5150
Influencer
Posts: 10
Liked: never
Joined: Dec 05, 2009 7:27 am
Full Name: Carl
Contact:

Re: Veeam Server Location

Post by ckd5150 »

But you'd be shifting from SAN based access of the VM in a push scenario to network based pull. either way the data ends up at the ESXi server. With our WAN link being 10G (and an open support case regarding the drop in speed with ESXi), i don't see how speed would improve.

Vitaliy S.
Product Manager
Posts: 24243
Liked: 1859 times
Joined: Mar 30, 2009 9:13 am
Full Name: Vitaliy Safarov
Contact:

Re: Veeam Server Location

Post by Vitaliy S. »

Hi Charles,

Take a look at this post from Anton, should make it more clear.

Furthermore, here is a real life replication testing example from Enrico. Those figures illustrate performance rates difference between placing backup server onsite or offsite.

Thanks.

whynotq
Influencer
Posts: 12
Liked: never
Joined: Mar 24, 2010 2:29 pm
Full Name: paul whyton
Location: UK, South
Contact:

Replication Push or Pull?

Post by whynotq »

[merged]

We have a situation where testing of the performance differences between Pushing replication data and Pulling replication data isn't possible as the Pull operation fails with this error:

Initializing target session
PrepStorageForWriteEx failed 'nfc://conn:fcs-dr-vc1,nfchost:host-126,stg:datastore-130@VeeamBackup/TestServer(vm-969)/2011-07-06T103747.vbk'
Client error: NFC storage connection is unavailable. Storage: [stg:datastore-130,nfchost:host-126,conn:fcs-dr-vc1]. Storage display name: [esxdr1_datastore2].
Failed to open NFC storage [nfc://conn:fcs-dr-vc1,nfchost:host-126,stg:datastore-130@VeeamBackup/TestServer(vm-969)/2011-07-06T103747.vbk] for read/write access

however the test "Push" job to the same target storage works fine. I've had a search of the forum and see the error relating to a timeout issue or possible connection issue but as the job is the same VM and same target over the same wire why is it failing the way it is? is it genuinely a timeout issue from the remote site host?

Vitaliy S.
Product Manager
Posts: 24243
Liked: 1859 times
Joined: Mar 30, 2009 9:13 am
Full Name: Vitaliy Safarov
Contact:

Re: Replication Push or Pull?

Post by Vitaliy S. »

Basically this message means that we face timeout issues (for ex. due to high latency) from the main site to the DR one while trying to update and write to target storage files.

This can mainly happen because of the agents that update VBK, VRB and Virtual Machine's virtual disks files are started on the main site, which makes all this process a bit complex, cause you'll have additional traffic flowing from the source to target site and vice versa.

Gostev
SVP, Product Management
Posts: 26679
Liked: 4268 times
Joined: Jan 01, 2006 1:01 am
Location: Baar, Switzerland
Contact:

Re: Veeam Server Location

Post by Gostev »

Yep, with v5 and ESXi target, which I believe is what used here, pull is really the only good option for the reasons referenced above. Also, consider that with backup server in DR site, you will still have it available for performing failover in case of disaster in production site (good thing!)

acomiskey
Novice
Posts: 4
Liked: 1 time
Joined: Aug 26, 2011 7:20 pm
Full Name: Adam Comiskey

Replication questions

Post by acomiskey »

[merged]

Currently running version 5.

Ran a local replication of about a dozen vm's. Took the replica server to our dr site, changed the ip address, and retried the replication job. We have about a 24Mb pipe between sites.

Replication job is only using about 8-10Mbps to the dr site, and for some reason about 6-8Mbps coming back from the dr site. Why is this? Why is there this much data coming back from the replica server? Seems to me like something is wrong and it seems awfully slow. I tested by replicating a single windows machine using changed block tracking. Ran 2 replications back to back. Although the second replication was faster than the first, it still took 20 minutes, and only used about 10Mbps. I can't imagine anything changed much on the server between the 2 replications. Does this sound normal?

Post Reply

Who is online

Users browsing this forum: No registered users and 17 guests