An agentless architecture doesn't require the management of agent software; however, it must rely on established communication protocols (e.g. FTP, CIFS/SMB, TELNET, SSH) to communicate with the client. Therefore, the backup server must communicate with each client using a pre-arranged username and password, usually a privileged one. (It must be privileged enough to access all data that must be backed up.)
I think either architecture would work just fine in an SMB environment. It's the enterprise that would have the strongest opinions about the values or merits of an agent-based or agentless architecture. While a company with 1000 systems to back up would see a great benefit from not having to manage an agent on each machine, they might also object to storing the username and password of a privileged account on the backup server (even if it's encrypted).
Do you know…
This was first published in April 2006