Comprehensive data protection for all workloads
Post Reply
leedudmish
Influencer
Posts: 11
Liked: 2 times
Joined: May 09, 2016 6:13 pm
Full Name: Lee
Contact:

Replication

Post by leedudmish »

Hi All,

We have been using Veeam backup for the past couple of years and are about to setup DR servers we will be replicating to as failover hosts.

Due to constraints with air con in the DR location (we are talking a listed building) the Veeam servers & Storage cannot go to the same site as the DR servers as I had planned originally to do a replicate from backup file locally on the switch at site to the DR servers, this then creates a full DR suite of backups off site along with failover servers and not impacting production at all.

We currently have 3 Veeam servers carrying out backups across our estate and these will each be replicating after the backup window has completed using the backup file as the replica.

We have a dedicated 1Gbps WAN link between offices so a max bandwidth of 100MB/Sec throughput however in theory the veeams will be under load during backup and we don't have quick storage for the veeam backups so I dont think we can reach that sort of speed off disk to run the reps overnight at the same time, our veeams use all night to carryout backups and only just finish in time for the morning production.

We potentially plan to therefore replicate during the day over the link with some sort of bandwidth throttling say 25% of max bandwidth available (I am running numbers to work out how much data we have changing and need to replicate). This would leave the other 75% free for production use, we have no QOS on the line at all as it is not a routed subnet (basically a point to point GB line) so it is a free for all however our bandwidth on the wan link is sporadic and usage is fits and bursts if I had to describe its usage.

I am just curious if anyone has ever done anything like this and if anyone could answer the questions below as it may alleviate some of the problems or maybe give some advice to help my case as to why we should not do it this way.

> What happens if a backup job becomes corrupt and you have to re run the job from scratch can the replica pickup or would it need fully replicating again.

> How does the replica deal with say a disconnection whilst the replication job is running - can it pick-up and be re-run or does it have to be started from scratch what happens if it cannot pickup do we need to re run the full replica?

> When firing up the replica after a disconnection - say we had a disaster at the main site half way through the replication jobs and the latest point was not completed can we fire up the previous day's replica point or would the chain be broken? - say we keep 3 points of each VM on the failover so can go back 3 days not that we would want to go that far back.

> Is there a way to have a second storage location which we could replicate from e.g. a large Nas box locally which is much smaller in terms of heat output than the three servers we have currently and then set-up the Veeam servers to replicate to the failover hosts from this storage so all the main data exchange is over a local 1GB switch rather than through the WAN link?

> How does the secondary destination work on the backups if the jobs go wrong if you re run the job due to an error is the secondary updated at the same time as that may impact the line if we had to re run.

> Assuming we had a setup where it was just the DR servers off site, I assume I can still go into the vmware hosts and start the vm's manually - I am thinking in a disaster we loose the veeam server at the main office also and we would be unable to initiate failover on the veeam console.

Ideally I would much prefer the veeams at the DR site as it also means we have the backup infrastructure to start backing up the newly fired up failover hosts aswell but due to the age of the building and restrictions imposed its not possible, we have a self contained air con unit but as a team we dont believe it will cope with the extra servers.

Any help/feedback/questions is appreciated.

thanks

Lee
PTide
Product Manager
Posts: 6408
Liked: 724 times
Joined: May 19, 2015 1:46 pm
Contact:

Re: Replication

Post by PTide »

Hi and welcome to the community,
How does the replica deal with say a disconnection whilst the replication job is running
Yes, there is a resume on disconnect feature.
What happens if a backup job becomes corrupt and you have to re run the job from scratch can the replica pickup or would it need fully replicating again.
It can pick up - there is a Replica Mapping mechanism for that.
<...>can we fire up the previous day's replica point or would the chain be broken?
Yes, you can spin up previous day's replica.
<...>Is there a way to have a second storage location which we could replicate from<...>?
Yes, there is a Replica from Backup fetaure that can be used in a conjunction with Backup Copy job.
How does the secondary destination work on the backups if the jobs go wrong
The secondary destination get's updated during sync intervals.
Assuming we had a setup where it was just the DR servers off site, I assume I can still go into the vmware hosts and start the vm's manually
Correct. You can even test that manually.

Please also review this thread for more information.

Thank you.
leedudmish
Influencer
Posts: 11
Liked: 2 times
Joined: May 09, 2016 6:13 pm
Full Name: Lee
Contact:

Re: Replication

Post by leedudmish »

Hi Ptide

thanks very much for the quick and precise responses will give the thread a read for more info

Lee
Post Reply

Who is online

Users browsing this forum: Bing [Bot] and 202 guests