This document provides a discussion of computer backup solutions and our recommendations.
For computer backups we generally distinguish between two types: file and image. A file backup is simply a copy, or snapshot, of a set of files at a point in time. An image backup is a complete snapshot of the system state. Generally, a file backup allows for recovery of one or many discrete files while an image backup allows for the complete restoration of the operating system, including installed software. The image backup generally does not support recovery of a single file or set of files; the entire image must be restored to the drive, potentially overwriting modified files taken between the time of the last image backup and restoration.
For most purposes, a file backup is sufficient. If you use specialized software packages that generally aren’t available from http://software.sites.unc.edu/software/ you should keep the original installation media and consult with the vendor’s documentation to ensure file backups include application configuration files for restoring settings and data in the event the application must be reinstalled on a new system.
A common example is Mozilla Firefox. One can use the export bookmarks function through the application to backup only the bookmarks for transferring to a new installation http://mzl.la/1BAQB3f. Alternatively one can include their Profile folder in a scheduled file backup to restore all settings, including bookmarks http://mzl.la/1xKrFsk.
Consult with itsop@unc.edu for assistance.
Managed
Managed solutions are provided automatically by hosted services and do not require additional configuration by you:
- myfiles.unc.edu (L drive), storage.unc.edu (M drive) and secnas (N drive) all provide the ‘previous version’ functionality. ITSOP recommends using these locations for critical data. For access to storage.unc.edu (M drive) and secnas (N drive) refer to the respective pages under the Services tab in the Navigation menu at the top of this page.
Unmanaged
These require opt-in and must be set up, configured by you:
- Windows Backup – http://windows.microsoft.com/en-us/windows7/products/features/backup-and-restore. Provides both full system image and file backup. This is an excellent complement to special purpose computers, such as those that interface with an instrument or HPLC hardware.
- Mac Time Machine – http://support.apple.com/en-us/HT201250. Provides both system image and file backup.
RAID is NOT a Backup
Increasingly systems are shipping with RAID (redundant array of inexpensive disks) and as a result ITSOP is seeing an increase in reports of RAID failure. RAID should not be considered a replacement for a backup; RAID only protects from disk failure, it doesn’t protect from other types of failure such as software crashes, malware, etc. Read on for our recommendations on how to implement a backup and how to regularly monitor the state of the RAID so that you’re not a victim of data loss.
Wikipedia has a well referenced discussion of RAID technology http://en.wikipedia.org/wiki/RAID. Wolfram Alpha also has a really useful calculator http://www.wolframalpha.com/input/?i=raid for understanding how a particular configuration works.
RAID1 and RAID5 are the most common configurations we’re seeing. RAID1 is two mirrored disks; one disk can fail without data loss. RAID5 involves three or more disks, where blocks of data are distributed across the disks; RAID5 also supports one disk failing without data loss.
When a disk fails, it should be replaced as soon as possible. The specifics of how to manage the RAID will vary between vendors and platforms; however, generally, the system should supply some form of disk verification, typically displayed when the computer is restarted prior to the operating system starting. If the notification indicates the RAID is operating in ‘degraded’ state it should specify which disk in the array has failed. Once the disk is replaced, the system should detect the new disk and attempt to reconstruct the array by writing data from the functioning disks. This can take several hours and depends on the size of the disk.
Note that the system will continue to operate even if the RAID is degraded, however a second disk failure will result in data loss. You should regularly check the RAID disk verification utility supplied with the system and respond to disk failure immediately: either by contacting the vendor for non-CCI computers or itsop@unc.edu for CCI computers. If the disk can be replaced under warranty, this is generally preferable though ITSOP can assist with replacing out of warranty equipment.
If you’d like assistance with reviewing the particular RAID configuration contact itsop@unc.edu for a consultation.