Should you backup all your data?
I was a backup admin for many years. Backing up and restoring data was my job (along with managing high-end arrays and server infrastructure), with the brief to backup anything and everything. No questions asked, because when advising what we should be doing to streamline backups, I was always fobbed off.
So file servers had every file backed up that belonged to users. This included illegal files like movies and music, as well as files and folders that hadn't been touched in months, or even years. This was a business requirement, but the question to ask is just this: should all data be backed up in an environment?
While tape continues to grow in capacity, along with disk which is also becoming cheaper per GB/TB as the years go on, the simple answer would be to do just that...backup everything without regard for backup times, and even more importantly, restore times. The implications behind this include, but are not limited too:
- Longer backup times with the amount of data being backed up. Lots of small files, especially when they might be excluded, add time to backups, resulting in growing backup windows. End result? Backups running outside of traditional backup windows!
- Much longer restore times as all these files would be written back to disk in the even of server failure (been there, done that and it can take days if you're still doing agent-based backups!).
- Inefficient usage of storage, both tape and disk. Wouldn't you want to save money by buying a smaller capacity tape drive/autoloader/library and using the excess cash elsewhere? Disk space might be at a premium and therefore having files on disk that shouldn't be there adds up at the end of the day.
The alternative answer, and something I ended up doing, was to not backup everything, and there are many benefits for doing your backups this way. Means to accomplish this would include excluding all these files from your backup. Backup Exec would simply skip the file/s when it encounters them meaning they would not be streamed off to tape. The downside to this would be that they would still exist on disk.
These files could be deleted from disk manually, especially if they're said illegal files.
For files that haven't been used in months/years, and are designated as being archivable, this would be an option as well...move to a different location (either archiving through Symantec's product) or copying/pasting manually, stream off to tape/disk and label accordingly. The files could then be deleted.
Archiving with specialised software would, in the long-run, be more beneficial as retention periods could be properly assigned, and data within a specific date range archived off automatically. For long-term retention to disk and/or tape, archiving is something worth investigating.
Not backing up all the data would ensure backup windows would be met, disk space cleared up, less capital outlay for new hardware to cater for this etc.
So...should you be backing up all your data? If not, it's time to start checking out your environment to see where you can become more efficient!