Why do restores take longer than backups?

The performance variation between both operations is mostly attributable to the number of steps involved in each. Essentially, a backup is not much more that a read operation. Data is simply read from the backup client and sent to the backup server.

Conversely, a restore process implies a write operation. Because data is usually being written to a file system during a restore operation, it involves some file system overhead such as creating an entry in a file allocation table or journal, permission setting, ACL and various other file attributes being recorded. This obviously takes more system cycles and ends up being much slower than a simple backup (read) operation.

Some file systems are better than others are at this and a significant restore performance difference can be noted. However, generally speaking, the worst possible restore performance will be observed when restoring a large number (millions) of small files. In comparison, restoring a 500 GB database will be dramatically faster that restoring a 500 GB file server.