• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Veeam Backup & Replication

J

jenuster

Limp Gawd
Joined
Mar 8, 2011
Messages
154
Learned something interesting today about how Veeam replication works:

When you failover to your disaster recovery site using Veeam, you cannot fail back to your production site--you have to actually sent the entire VM over the WAN link to the production site (I am assuming the production servers weren't destroyed). The current version of Veeam is not able to sent the incremental changes back to the production environment. Also, once you sent the replica over to the production site, you will need to delete all replicas at the disaster site and reseed the VM.



Maybe you guys knew this but this came as a surprise to me. But it does make sense in the fact that Veeam Replication is a disaster recovery software and not a high availability/business continuity solution.
 
To be clear so that I am following what you were testing here:

Production VM site goes down.
Failover to your DR VM site and everything is up and working. Production data is being put to your DR VM now.
Production VM site is restored to same place where it was before the failover.

Now what you are trying to do is get the data from your now 'production' VM that lives on your DR VM site to the production site. The only way you are seeing to do this is to do a full 'backup' of the new production at the DR site to the original production site. Get that running and do another full backup to your DR site but having to delete all the old backups for the VM to continue to do inc/diff backups.


Assuming the above is correct, it doesn't seem right. I understand having to do the full from the DR to production. I would think that you would be able to use that full for your 'full' on the DR site and not have to do yet another full from Production to the DR.
 
That's correct. I honestly don't understand why you couldn't just sent the incrementals over once your production site is back up--assuming it wasn't a total disaster.

After you do a full from the DR to prod, you need to delete all the replicas in the DR site and then reseed everything (basically setting everything up from scratch)

I talked to a Veeam Lead tech today and that's what he told me.
 
Working with replications and Veeam is next on my to do list. Looking at setting up a DR site with www.iland.com

From what it sounds like your stuck wiping the production side either way in a DR event. I was thinking fail over to DR, once production site is ready replicate back from DR to production while DR site is running. Once first replication is done, assuming first replication takes hours over WAN link, power down DR and do a final replication back to production.

Wipe the DR and start over with replication jobs from production to DR as originally setup.

Assuming you can take the downtime with the last replication to grab the last of the changes. If you can't then I'm guessing the design is wrong with the wrong product. Need to ratchet things up to vmotion/vstorage/FT, HA within the services offered on the servers and big WAN pipes. Over my head, true enterprise setups.
 
"
Once first replication is done, assuming first replication takes hours over WAN link, power down DR and do a final replication back to production.

"
By final replication back to production, I think you mean sync the incremental changes from DR to prod--that's where the problem lies. You can't sync incremental changes from DR to Prod--you have to replicate the ENTIRE VM.

Let me know if that makes sense.
 
Hey everyone, time for a bump! i got a new job and being tasked with learning how to use Veeam. I am going to be setting up the free trial this weekend in my home lab (if i am not lazy) lol.

A few questions that have me puzzled:

Is the replication real time as changes are made to the VM or does replication only happen when it is scheduled to do so like a backup? Basically I am wondering what the difference is between a backup and replication of a VM? i assume that replication is like constant incremental backup?

I was reading through the Veeam manual for the Backup and Replication program and everything seems pretty simple but I am not sure how the failover works. Does it work where if the VM at one site goes down the replicated VM at the other site will turn on and start acting as the main VM? Is this just like VMware HA?

This is what the manual says and it sounds just like VMware HA?

Failover
In order to diminish the risk of failure and make the work process seamless, Veeam Backup & Replication 5.0 provides a possibility to failover a virtual machine to its replicated version. In case of software or hardware malfunction, you can recover a corrupted virtual machine by failing over to its replica or its last known good point-in-time incremental.
At replica failover, the Veeam Agent is started on the target ESX server where a replicated virtual machine resides. A snapshot of a replica is created to protect a replicated VM from user's changes, and the replica is started on the target host.
In case the undo failover operation is performed, the replica reverts to the created snapshot. Any changes made to a replicated VM will not be committed to the original VM when undo failover operation is performed.
At performing failover, the original VM should be stopped.
Click to expand...

Also, i was just wondering how this program can do the file level backup and restore. I just learned how to use Backup Exec and it needs the remote agent on each machine to do this. With Veeam it does what Backup Exec does but with only an agent on each ESXi host. how does it do such a thing?
 
Last edited:
Use SRM for DR instead (especially 5.0). WAY better, if your arrays support replication :)
 
Thanks for the info, I will look into SRM but from a brief google search it looks like it will cost a lot.

Now can somebody explain to me the difference between replication and a backup?:confused:
 
SRM is likely too expensive if you have a small environment and don't need or can't justify the cost to have automated failover.

SAN level replication or something like Veeam is the most likely balance between cost and complexity.
 
hutchingsp said:
SRM is likely too expensive if you have a small environment and don't need or can't justify the cost to have automated failover.

SAN level replication or something like Veeam is the most likely balance between cost and complexity.
Click to expand...

Keep in mind that licensing changed on SRM 5. If you're protecting <75 VMs it's $195/VM (list price..street is lower). That includes the host based replication piece too.
 
AMD_Gamer said:
Thanks for the info, I will look into SRM but from a brief google search it looks like it will cost a lot.

Now can somebody explain to me the difference between replication and a backup?:confused:
Click to expand...

If I replicate data it means I have two copies..usually close to being in sync. If I corrupt my primary data that corruption will then be replicated to the other side. That's why you still do backups...or snapshots...or something to protect against that so if you need to roll back a week you can.

Veeam blurs it a bit. They do backup. They do replication. They also let you boot VMs off that backed up and replicated data for really fast recovery. Great product..you just need to first set your requirements for how, when, and where you want to recover from different failures.
 
NetJunkie said:
Keep in mind that licensing changed on SRM 5. If you're protecting <75 VMs it's $195/VM (list price..street is lower). That includes the host based replication piece too.
Click to expand...


NetJunkie said:
If I replicate data it means I have two copies..usually close to being in sync. If I corrupt my primary data that corruption will then be replicated to the other side. That's why you still do backups...or snapshots...or something to protect against that so if you need to roll back a week you can.

Veeam blurs it a bit. They do backup. They do replication. They also let you boot VMs off that backed up and replicated data for really fast recovery. Great product..you just need to first set your requirements for how, when, and where you want to recover from different failures.
Click to expand...

Both of these, and SRM lets you do a lot more now in 5.0 as well. Replication is for DR, comets, asteroids, hurricanes, etc, backups are for lost vms/corruption/oops I deleted something. Very different purposes. :)
 
NetJunkie said:
Keep in mind that licensing changed on SRM 5. If you're protecting <75 VMs it's $195/VM (list price..street is lower). That includes the host based replication piece too.
Click to expand...

Thanks :) I hadn't realised they'd moved to per-VM pricing for that too. That gives some very interesting possibilities for smaller shops with a low VM count!
 
They went per-VM a while back on SRM 4. With SRM 5 they split licensing to two levels.
 
And I have to admit, the product really does rock (my bias aside) - once configured (a bit more difficult than it used to be), it's infinitely easier to manage and operate, and lets you do a LOT more than you could before. A true second generation product.
 
Any experience with SRM and Host Based Replication (HBR) over ~ 10 Mbps WAN connections? For companies w/o SAN's.

Any hosted cloud based SRM/HBR providers one can recommend?

Been trying to make a Veeam solution work over the WAN and it's just not going.
 
Required bandwidth depends completely on your write percentage/amount and your RPO.
 
NetJunkie said:
Required bandwidth depends completely on your write percentage/amount and your RPO.
Click to expand...

Duration aside, can you speak to reliability of VMware's SRM across the WAN using HBR/non SAN setups?
 
marty9876 said:
Duration aside, can you speak to reliability of VMware's SRM across the WAN using HBR/non SAN setups?
Click to expand...

No one really knows. SRM 5 has only been out like 2 weeks. We don't have anyone doing HBR yet. Coming...but not yet.
 
NetJunkie said:
No one really knows. SRM 5 has only been out like 2 weeks. We don't have anyone doing HBR yet. Coming...but not yet.
Click to expand...

Ok fine, I'll wait another week! :p

Thanks for the feedback. Looks like SRM/HBR would need VMware Standard licensing plus the SRM add on'sn which would be a big jump for me with coming from VMware Essentials.
 
Yeah, no idea on the network reliability yet :p I'll play with it more when I'm back (getting married/honeymoon these two weeks). :)
 
Boo, a bit disappointed by the fact that SRM 5 using HBR doesn't support failover...guess it makes sense.

Looks like I'll continue using VEEAM replication.
 
jenuster said:
Boo, a bit disappointed by the fact that SRM 5 using HBR doesn't support failover...guess it makes sense.

Looks like I'll continue using VEEAM replication.
Click to expand...

it doesn't support failBACK. It fully supports failover, and failback by hand, but not automated failback :)
 
lopoetve said:
it doesn't support failBACK. It fully supports failover, and failback by hand, but not automated failback :)
Click to expand...

Yeah..SRM without failover is....well..it's not anything. It doesn't do automated failback which is a feature in SRM5.
 
I actually meant fail back by the way. By "By hand" do you mean manually copying the VM and VMDK's from the disaster site, driving/flying to the production site and importing the VM?
 
jenuster said:
I actually meant fail back by the way. By "By hand" do you mean manually copying the VM and VMDK's from the disaster site, driving/flying to the production site and importing the VM?
Click to expand...

No... What we mean is that you'll have to reverse replication and reconfigure it before you can fail back. With an array it's basically, click a button and it does all that for you. SRM4 was manual failback as well but some vendors, like EMC, built their own failback automation tools.

So it's not a big deal..just something you don't get.
 
Fair enough. I'll test it out in the lab and see what happens. Also just to be clear, I am referring to SRM using Host Based Replication. So Site A I have a storage array and Site B I am using local storage. Definitely am interested to learn about failing back a particular VM from Site B to Site A in this scenario.
 
Last edited:
You simply walk through all the steps that you had to do to configure it for A->B, only this time from B->A. :) Instead of a button.
 
Got it! Thanks for your help. By the way Veeam Backup and replication 6 is going to finally support Failback. It's supposed to come out in the 4th quarter.
 
For a small setup like 2-3 ESXi hosts is it ok to have Veeam B&R on the same VM as vCenter Server? I was reading it is only an issue if you are managing a large number of ESXi hosts from that vCenter. What do you guys think?
 
That should work, although it's not best practice. I don't think you will be able to do a Veeam Backup of Virtual Center though.

If you are using iscsi, I would set up a physical machine configured on the iscsi switch. This makes the backups faster and also won't clog your LAN when the backups are taking place.
 
I use a stand-alone physical server as my Veeam B&R. Backup is one thing, but think about recovery. If the shit hits the fan on my ESXi cluster and associated storage I can stand up the production VMs on a totally separate system in minutes and they can limp along on that server till the ESXi cluster is back in business.

Running Veeam on the infrastructure that Veeam is supposed to protect against is ill advised imho.
 
I am confused how the replication failover works with networking, say you are replicating a web server how does the outside world get to it once you do the failover to the replica at a different site?

bump.

My remote site uses a different IP subnet so what happens when the replica powers on?
 
Last edited:
at least in SRM, IP addresses/etc can be changed, and switch routing tables could be scripted through the command callouts.
 
You must log in or register to reply here.
Back
Top