16.1. Introduction

Making good backups is a fundamental chore, and one that is too often messy and inconvenient. This chapter covers using rsync and Mondo Rescue for robust, easy backups and, just as important, easy restores.

rsync is extremely efficient. It only transfers changes in files, and it can perform on-the-fly compression. It excels at keeping file trees sychronized; unlike a lot of backup software, it even mirrors deletions. Because of these features, rsync is the tool of choice for updating and mirroring web sites, CVS trees, and other large, complex file trees.

There are two ways to use rsync: over ssh, for authenticated login and transport; or by running it as a daemon to create public archives, which do not use ssh. Using ssh requires users to have login accounts on every machine for which they need rsync access. When rsync is run in daemon mode, you can use its built-in authentication methods to control access, so that users do not need login accounts on the rsync server.

Mondo Rescue is a great utility for creating a bootable system restore disk. Use it to make a snapshot of your current system configuration, and a bare-metal rebuild is as easy as booting to a CD or DVD. You'll have all of your patches, configurations, and updates. Then pull your data from your rsync server, and you're back in business in record time. Mondo can also be used for a complete system backup, all by itself.

Tape backups have long been a mainstay, because they used to provide the most storage for the least money. But they are slow to write to, and even slower to restore from. If you're getting paid by the hour, it might not seem so bad to sit back and enjoy the endless whirring when you're looking for a little bitty text file at the end of the tape. But there are other disadvantages:

• Hard drives are much larger than tapes now, so you may have to use a tape autoloader, which is expensive, or have a designated tape-changing minion, which can raise dependability issues.

• Users must depend on system administrators to restore files.

• Doing a bare-metal restore is complicated, because you must first install a filesystem, then the tape drivers, and then do the restore.

• Even when a single tape is large enough to store a complete backup, you still have to change and store tapes, even if it's only on a weekly basis.

Tape has one advantage: longevity. CDs, DVDs, and hard drives cannot be counted on for long-term archiving. I wouldn't count on home-burned CDs/DVDs for longer than two years. Tapes should be good for 20 years, if stored carefully. Of course, there's no guarantee that there will be drives or software to read them in 20 years, but the tapes themselves should last just fine.

Longevity is the primary difficulty with all digital storage media, because even if the medium (tape, DVD, etc.) survives, there's no guarantee that the tools to read it will endure. Hardware is in a continual state of change, and file formats change too. Can you still read 5.25" floppy disks? WordStar files? VisiCalc? Even the most offbeat obsolete file can be read, if it's a plain ASCII file. But what are you going to do with proprietary binary datafile formats, like .ppt and .pst? If the vendor decides to stop supporting them, or goes out of business, you're out of luck.

Paper is still the long-term storage champion. (Yes, stone tablets, parchment scrolls, and clay tablets win hands-down for sheer durability, but let's stick to contemporary media.) However, movies and songs do not translate well to paper, and paper is hundreds of times bulkier than any digital storage medium.

So, as a practical matter, planning good short-term archiving and disaster recovery is usually the best we can do. rsync and Mondo Rescue are perfect for this. rsync is fast, compact, and completely automatic no need to remember to swap discs or tapes. Restores are easy and fast. You can even set up user-accessible backup archives, so users don't have to pester you to restore files.

What about backing up your backup server? No problem. Setting up a remote rsync mirror for backing up the backups is a common strategy.

Also, there are more choices than ever for removable media, including removable drive trays for IDE drives, USB/FireWire portable drives, and flash storage devices. In sum, there are a lot of excellent options for designing your backup and recovery schemes.