Our various network services - the Netware cluster (Causeway) which provides the personal, departmental and course drive space, Groupwise and Netmail email servers, our Microsoft SQL Server, our web servers, the Library's catalog server, Blackboard, and so forth all consume a substantial amount of disk space. Backing up that critical data is difficult and expensive. Generally, a business or organization has to set goals and priorities - what data and system information is important enough to back up, how frequently it needs to be backed up, and how much recovery time is acceptable. An organization could decide that certain information is not important enough to back up, or to back up often - the risk or consequences of losing that particular data or system is not significant enough to justify the cost of backing up that data or system.
At this time, we are backing up about a terabyte of data and critical system files. As a side note, that is by no means everything. Due to resource limitations (equipment, bandwidth, ability to schedule backups during non-peak times), we have had to pick some non-critical systems that do not contain data and either back them up less often, or skip them knowing we can recreate the system from scratch in a day or two.
A backup is not a perfect solution: it involves moving a tremendous amount of data. Everything has to work correctly - the server managing the backups, the server being backed up, the network, our storage network, our tape libraries and the hardware use to connect them to our storage network all have to function for the duration of a backup. As an example, it currently takes about seven hours to back up the entire disk volume containing the user directories (your F: drive). If anything goes wrong with part of the process during those seven hours, the backup can fail and we have to start over.
It is not practical with traditional file-based backups to do that every night. On the large volumes - Users, Depts, Courses, the Groupwise post offices and the old Netmail server's email volume, we do a full backup every week, either overnight on on the weekend. Other nights, we perform either a *differential* or *incremental* backup - that is, only files that have been changed since the last full (or "base") backup are copied to tape with differential backups, or since the last incremental in the case of that type. Therefore, to reconstruct a volume, we would need to restore files from the latest full / base backup, plus the most recent differential backup or all the incremental backups since the last full / base backup.
Some things won't be backed up, though. For example, if a file is created during the business day but is deleted before the nightly user volume backup starts at 10pm, it is never going to be put on a tape. With our current resources, there is nothing we can do to address that. However, there is the salvage feature provided by Novell Netware which allows end users to recover deleted files in many situations.