What is a backup proxy?
A backup proxy is the component of a backup system that does the data-moving work. It reads data from the source, such as a hypervisor datastore, compresses and deduplicates it, optionally encrypts it, and sends it to the backup repository. The backup server coordinates jobs, and proxies carry the data.
What does a backup proxy do during a job?
When a job starts, the backup server assigns the workload to a proxy. The proxy retrieves the changed blocks, usually identified through changed block tracking, applies data reduction and encryption, and transfers the result to the repository or to a gateway in front of it. During a restore the same component works in reverse, writing data back to the production storage.
What are transport modes?
A transport mode is the path a proxy uses to read source data. The choice affects speed and the load placed on the production network.
| Mode | How data is read | Trade-off |
|---|---|---|
| Direct storage access | Straight from the storage array or datastore | Fastest, needs storage connectivity |
| Virtual appliance (hot-add) | Disks attached to a proxy virtual machine | Fast, proxy must run on the same cluster |
| Network | Through the hypervisor host over the network | Works everywhere, limited by the network |
How does the proxy affect backup performance?
Each proxy can process a limited number of disks at once, set by its processor cores and memory. Too few proxies and jobs queue, which stretches the backup window. Too many concurrent tasks and the source storage or the repository becomes the limit instead. Job statistics usually show which stage is the bottleneck. Tuning job concurrency covers how to balance the stages.
Where should backup proxies be placed?
Proxies belong close to the data they read, so that the heavy, uncompressed transfer is short and only reduced data crosses longer links to the repository. In multi-site setups that means a proxy at each site. From a security point of view a proxy holds credentials for the hypervisor and the repository, so it should be patched and access to it restricted like any other backup component.
Frequently asked questions
Is the backup proxy the same as the backup server?
In small installations one machine does both jobs. In larger ones the server schedules and tracks jobs, and separate proxies move the data.
How many backup proxies are needed?
Enough to process the daily change within the backup window. Vendors publish sizing guidance based on cores per concurrent task.
Does a backup proxy store backup data?
No. It processes data in transit. Restore points are stored in the backup repository.