Appliance
The backup appliance orchestrates backup operations. This is where you configure credentials, workers and backup repository storage accounts, snapshot and backup policies, and can mange restore operations. As a control plane, the actual tasks are performed with workers. The appliance communicates with workers via a dedicated service bus.
Recommendations
If you want to change the appliance size you have to do this after the appliance has been deployed.
The following recommendations and examples apply to the latest VBAZ builds 1.
| Appliance size | Workloads to back up | Workers | Retention points |
|---|---|---|---|
| B2s (default) (2 vCPU, 4 GiB RAM with 32 GB premium SSD disk) | 300 | 50 | 17,500 |
| D4s_v3 (medium) (4 vCPU, 16 GiB RAM with 64 GB premium SSD disk) | 2000 | 250 | 90,0001(Veeam) |
| D8s_v3 (large) (8 vCPU, 32 GiB RAM with 128 GB premium SSD disk) | 3000 | 400 | 135,0001(Veeam) |
In order to properly function, there are 3 parameters that define the appliance size (and more important, the amount of memory required).
- OS utilization: the average value of memory needed in idle state which is around 1.5GiB
- VBA utilization: 5% of the total appliance memory required for the VBA application (Web UI and REST API service)
- Policies utilization: memory used per running policy (10MiB per policy and 1 MiB per additional workload in the policy). For more information on policy maximums, see the policies page
By combining all the above, the appliance size can be defined using the formula below which will calculate how much memory is required to operate.
OS utilization + VBA utilization + policies utilization
The below table contains some examples regarding the sizing of the appliance. The following applies to the table.
- Working space: amount of memory free after calculation of the amount required for the operating system and VBA application
- Workloads in policy: amount of workloads per policy
- Number of policies: amount of policies you can run within the working space (recommended)
- Memory consumed by all policies: total memory consumed by running all policies at the same time
Sizing examples
The below examples show some examples based upon a defined number of policies compared against the total appliance memory. This table is the maximum of memory usage when all policies would run at the same time.
In the table, we will use 10 MiB as a base per policy presuming that each policy has a snapshot and backup schedule configured.
| Total Appliance Memory | Working Space | Number of Policies | Workloads per Policy | Total Memory Consumed |
|---|---|---|---|---|
| 4 GiB | 4 GiB - 1.5 GiB - 4 GiB * 0.05 = 2.3 GiB | 50 | 25 | (10 + (25 * 1)) * 50 = ~1.8 GiB |
| 8 GiB | 8 GiB- 1.5 GiB - 8 GiB * 0.05 = 6.1 GiB | 50 | 75 | (10 + (75 * 1)) * 50 = ~4.3 GiB |
| 8 GiB | 8 GiB- 1.5 GiB - 8 GiB * 0.05 = 6.1 GiB | 50 | 100 | (10 + (100 * 1)) * 50 = ~5.5 GiB |
| 16 GiB | 16 GiB - 1.5 GiB - 16 GiB * 0.05 = 13.7 GiB | 200 | 50 | (10 + (50 * 1)) * 200 = ~12 GiB |
| 32 GiB | 32 GiB- 1.5 GiB - 32 GiB * 0.05 = 28.9 GiB | 300 | 75 | (10 + (75 * 1)) * 300> = ~25.5 GiB |
The main aspect of which will define how much memory you require for the appliance is how many workloads you want to configure within a specific policy. Veeam recommends to not size the appliance to the minimal limit but leave an additional margin of 20% RAM.
NOTE: The above information is only related to the OS, Veeam services and assurance to operate without issues. It does not take into calculations any additional software that may needs to be installed by company policies.
VBR integration
When integrating VBAZ with VBR, policy and retention data is imported into the VBR database.
You can connect more than one VBAZ appliance to a single VBR server, however make sure you size your VBR infrastructure appropriately. If working on a cross-geography cloud project it makes sense to use a VBR server per geo-region, to alleviate issues relating to latency as well and help with meet potential data residency regulations.