Replication Better automatically Job

[tovonantenaina]

Hi All,


I d'like to put a replication from my VMs from primary Server to the secondary server and we have 20 mbps of interconnectivity between two locations .

I would like to know the standart configuration for replication Job please ?

because for one of my VM , there is lot of trasancation (insertion in database ) .

So I put an automatically Job , scheduled perodically every 1 minutes

And Automatic retry :
- Retry times processing : 3 times
- Wait before each retry attemps for : 10 minutes

What Is the a better option ? or is there some one have any experiences about it ?


BR
Tovo

[Mildur]

Hi Tovo

There is no "standard". There is a business requirement from your company and it's called Recovery Point Objective (RPO).
RPO defines the maximum amount of data (time value) your company can loose.
Example: If you have an RPO of 1 hour, you need a replica at least every one hour to fulfill this requirement.

What's your expected RPO? 1 minute is probably too low and won't work with just 20mbps and snapshot creation for a database server.

Best,
Fabian

[tovonantenaina]

Hi ,

Our RPO is arround 15 mn but we except to put 1 mn , to be sure that the size of snapshot will be the minimal possible , in order to be transfert by this 20 Mbps interconnectivity .
What is for you the best way if we have a low Bandwith like we have

BR
Tovo

[Mildur]

Hi Tovo

1 minute will already be an issue for creating the snapshot, replicate the data and then delete the snapshot. Also 1 minute replications could lead to more data transfer than doing it all 15 minutes.
Example: A user changes the exact same 1GB File inside the VM every minute. 1GB changed blocks needs to be replicated each minute. If you have set replication to 15min, this 1GB would only be replicated once each 15 minutes.
- 1 minute replicas: 15 GB data transferred in 15 minutes
- 15 minutes replicas: 1 GB data transferred in 15 minutes

To minimize data transfer, you should check out WAN Accelerator:
https://helpcenter.veeam.com/docs/backu ... ml?ver=110

But I still expect 1 min as undoable because of the snapshots. I would start testing with 15min. If it took longer than 15min, deploy WAN Accelerator in your primary and secondary side as documented in our user guide.
If WAN Accelerator doesn't bring optimization, you need to organize faster network connection between both sides.

Thanks
Fabian