Backup software

Backup software is the application that schedules backup jobs, copies data from protected systems to separate storage, records each restore point in a catalog and runs restores on request. Its server and console form the control plane: whoever holds administrator rights there can change retention or delete restore points.

How is the work divided between control and data?

Backup software has two layers. The control plane decides what happens. The data path moves the bytes.

  • Control plane. A central server holds policies, schedules and credentials for the protected systems, and starts every job. A catalog records each restore point and where its data sits. The console is where administrators set retention, approve restores and delete restore points.
  • Data path. Proxies or agents read data from the protected systems, often reduce it with compression and deduplication, optionally encrypt it, and write it to a storage target.

What happens during a backup job?

The server starts the job at its scheduled time and tells the proxies or agents which sources to read. They read the changed data, process it and send it to the target. When the write completes, the server adds the new restore point to the catalog and applies retention, removing restore points that have aged out. A restore runs the same path in reverse, starting with a catalog lookup to find which pieces make up the chosen point.

Why does the console carry so much authority?

Every destructive action in the system is a normal console function. Shortening retention, deleting restore points and disabling jobs are housekeeping tasks, and in the job log they look the same whether an administrator or an intruder performed them. The server also stores credentials for hypervisors, databases and storage. Attackers therefore look for backup administrator logins early in an intrusion, before they encrypt anything.

How is the control plane separated from production?

Many organizations treat the backup server as a tier-zero system, on a par with the identity service. It sits outside the production domain so a stolen domain administrator password does not reach it, the console requires multi-factor authentication, and sign-in is limited to named people and locations. These measures narrow the administrative blast radius, the set of systems one stolen login can reach.

Separation has a limit. If retention exists only as a console setting, a login that gets past every control can still shorten it. Storage that enforces its own lock, such as S3 Object Lock in compliance mode, makes retention a property of the stored data, so a compromised console cannot delete what the storage refuses to release.

Frequently asked questions

Is backup software the same as backup storage?

No. The software manages jobs, catalogs and restores. The storage target holds the copies, and the two fail and are secured in different ways.

What is the difference between a proxy and an agent?

An agent runs inside the protected machine. A proxy reads data from outside it, for example through the hypervisor, so nothing is installed in each guest.

Does the backup software need its own backup?

Yes. Its configuration and catalog record what exists and how to open it, so they are kept separately from the repository they describe.