This section is intended for professionals who search for a best practice answer to sizing-related issues, and assumes you have already read the whole Veeam Backup for Nutanix AHV User Guide.
Be aware that a best practice is not the only answer available. It will fit in the majority of cases but can also be totally wrong under different circumstances. Make sure you understand the implications of the recommended practices, or request assistance. If in doubt, reach out to Veeam professionals on Veeam R&D Forums.
While allocating compute resources to the Nutanix AHV backup appliance, consider the following recommendations:
Number of VMs in Cluster
Number of Jobs
Resources for Management Tasks
Resources for Embedded Worker
No additional resources for management tasks
1 vCPU and 1 GB RAM for each concurrent task
1 vCPU, 1 GB RAM
It is not recommended to use the embedded worker for large deployments. Disable the embedded worker and use dedicated workers instead.
2 vCPU, 2 GB RAM
If you plan to use dedicated workers, it is recommended that you modify the backup appliance settings as follows:
- Set the maximum number of concurrent tasks to 0.
- Allocate resources to the backup appliance that are sufficient for management tasks.
To learn how to modify the appliance settings, see section Editing Backup Appliance.
While adding a dedicated worker to the backup infrastructure, consider the following:
- It is recommended that the number of configured workers does not exceed the number of hosts in the Nutanix AHV cluster.
- Each worker must be provided with sufficient compute resources to handle backup and restore tasks in parallel. The maximum number of concurrent tasks is configured in worker settings — if this number is exceeded, the worker will not start a new task until one of the current tasks finishes.
- It is recommended the total number of concurrent tasks configured for all workers deployed in the cluster does not exceed the number of physical disks added to the cluster. You can change the maximum number of concurrent tasks (the best practice is to allocate 1 vCPU and 1 GB RAM for each additional task) while deploying a new worker or editing settings of an existing one.