Seeding Replicas
This article covers some considerations and challenges with regard to seeding replica VMs
Last updated
This article covers some considerations and challenges with regard to seeding replica VMs
Last updated
Replication of extremely large VMs to cloud can involve unique challenges and require multiple attempts of different methods to ingest them into cloud storage due to the variables involved.
Hosted Network will assist in any way we can to ensure the process is as smooth as possible, however please be prepared to work with us through such challenges if required
By default, a replica job will replicate the entire source VM from the source infrastructure to the cloud infrastructure.
In most cases, unless bandwidth on the source side is very limited this is not an issue, however in the case of extremely large VMs or bandwidth limitations this method may not be viable. It is important to be aware of the following:
While the source Veeam B&R server is running a replica job, the live VMs being replicated are locked and not able to be backed up by a Veeam backup job. This can be an issue for seeding very large replicas when the source VM backups need to be able to continue while replica data is seeded
Uploading of large backup copies can take considerable time, especially if bandwidth is throttled on the source side in an effort to reduce disruption to the on-prem network
The longer a copy job runs for, the higher the likelihood something can go wrong (either on the source or destination side) and disrupt the job. This can often result in the job needing to start over from scratch
Upload speed available into the cloud repository may vary depending on the time of day and network load, and is not guaranteed to always be consistent or able to utilise the full upload capacity of the source connection
Seeding a replica from an existing on-prem backup is an alternative to direct replication and removes the need for Veeam to lock the live VMs while it completes the initial upload, by uploading this data from the on-prem Veeam backup and then syncing delta changes from the live VM when required.
This can be a viable alternative to direct replication for large VMs but has the following considerations:
An up to date and working backup of the VMs in the replica job needs to be available in the on-prem backup repository selected for seeding, otherwise the VM will still be uploaded using direct replication
This method does not address bandwidth limitations as the full volume of data is still uploaded from the source infrastructure to the cloud infrastructure
You must enable Replica Seeding on the first page of the replication job configuration and select the source repository on the seeding page
All backups for VMs included in the replica job need to be on the same repository as Veeam can only use seeding repository per job. If you have multiple backup repositories, you need to use separate replica jobs for the VMs on each
In cases where bandwidth limitations restrict the ability to upload replicas directly to cloud, it is possible to map a replica job to an existing cloud VM that is restored from a backup manually by Hosted Network.
This process requires the following high level steps:
VMs requiring replication are backed up using Veeam backup jobs (or backup copy jobs) to portable storage such as a NAS or USB drives
Storage is shipped to Hosted Network
Hosted Network restore the backups into VMs on our cloud infrastructure and import them into the appropriate tenancy
Replica job is created on the on-prem Veeam B&R server and the replica mapping feature used to connect the job to the existing cloud VMs
The replica job will then run a disk comparison between the source and destination VMs and replica only the required data to update the VM
This process involves similar steps and considerations as manual seeding of backup copies, please see Seeding Backup Copies in the BaaS section for more information