Auto-scaling is critical for backup and recovery operations, especially for large-scale operations. Auto-scaling allocates resources dynamically, based on demand, preventing resource bottlenecks and optimizing performance. It then decommissions access nodes when they're not needed. When you need to use auto recovery in a disaster scenario, where the volume of data and the number of workloads to restore can vary significantly, auto-scaling ensures that the necessary computing resources are available, without provisioning more than you need.
Supported Operations
Auto-scaling for is supported for most data protection operations, including file indexed backups. Only restores of guest files and instance files don't use auto-scaling.
Number of Access Nodes That Are Created
The number of access nodes that are required increases or decreases based on the amount of data you back up and the settings you configure for automatic scaling. By default, an unix access node is used for backup and restore operations.
For Backups
For backups, the software determines the number and size of access nodes (if not explicitly set in the scale profile) to create based on the following characteristics:
-
The Recovery Point Objective (RPO)
-
The throughput rate
-
The number of disks being backed up
-
The Maximum number of access nodes setting in the scale configuration
-
Required resources such as CPU, RAM, and disk space vary based on the job type (backup, restore, or file indexing)
Important
To increase the number of access nodes, change the Maximum number of access nodes setting instead of changing the number of data readers for each access node.
For Restores
For restores, the software determines the number of access nodes to create based on the following characteristics:
-
The number of VMs that are being restored
-
The load on auto-scaled access nodes
-
The Maximum number of access nodes setting in the scale configuration.
For File Indexing
For file indexing, the number of auto-scaled access nodes that are created is based on the number of CPUs for each access node and the number of streams for each CPU.
Operating System of Access Nodes That Are Created
For file indexing, the operating system of auto-scaled access nodes matches the OS of the VMs that are indexed.
Decommissioning of Access Nodes
Automatically commissioned access nodes are deleted from the cloud service provider portal if no active jobs are running across workloads for more than 10 minutes.
Log Files
The log files from dynamically deployed access node are streamed to the access node on which the job was initiated and are available at the following location:
StreamedLogs