W. Curtis Preston is the king of backups, and his book Backup and Recovery (BAR) is easily the best book available on the subject. Preston makes many good decisions in this book, covering open source projects and considerations for commercial solutions. Tool discussions are accompanied by sound advice and plenty of short war stories. If the author addresses the few concerns I have in his next edition, that should be a five star book.
The best aspect of BAR is the author's obvious expertise in this subject. He does a good job sharing lots of his knowledge with the reader. Probably the most valuable conceptual framework I learned in BAR is the difference between backups and archives. Pages 696-7 summarize this nicely: "Backups are the secondary copy of primary data... Archives are the primary copy of secondary data." In this section and elsewhere, Preston describes how archives are the repository one should create when answering ediscovery requests and similar queries -- not backups. This is an extremely powerful idea and I plan to see how my employer deals with this issue.
The second best aspect of BAR involves multiple chapters on backing up various databases. One can usually find similar coverage in single books on specific databases, but having all information in one book is useful for purposes of comparison. Chapter 15 provides an overview of the entire problem by discussing terminology and features found in many databases. This chapter helps storage admins understand the database admin world. Of particular note was the coverage of Microsoft Exchange, which the book calls a specialized database. I had not thought of Exchange in this light, but it's true -- especially when Microsoft indicates future versions will have SQL Server replacing Extensible Storage Engine. I only read chapters on SQL Server, Exchange, and MySQL.
The third best aspect of BAR includes OS-specific chapters on bare-metal recovery. Although my OS of choice (FreeBSD) didn't merit its own chapter, I felt the material in the bare-metal section was robust enough to help me perform this work if necessary. I really only read the chapters on Windows/Linux and ignored Solaris, HP-UX, AIX, and Mac OS X.
BAR is a good book, so why not five stars? First, I thought the chapters on open source backup options (especially ch 7 on "Open-Source Near CDP") were weak. I wanted to learn a lot more about rdiff-backup, for example, but the tool merited about 5 pages and introduced only the simplest possible invocation. Rsnapshot was also undercovered. It seemed like too many pages were spent on utilities I would probably never use (given newer options) like dump and cpio. I was also not confident I could get very far with Amanda, BackupPC, or Bacula given the detail given to each open source product. (Regarding BackupPC -- I had to guess it was open source and then only found out the truth when its Web site at sf.net was mentioned late in the chapter!)
Second, some topics never really made sense. For example, I still do not understand how snapshots actually work. Calling it a "picture" means nothing to me. Snapshots are mentioned throughout the text, and the explanation that finally appears near the end of the book in a miscellanea chapter doesn't help.
Third, I would really have liked to hear more about services offering backup to the Internet, like Amazon's S3 and others. This MUST be covered in the next edition.
Finally, although the book has lots of advice, it would have been nice to have had a case study chapter where multiple example enterprises demonstrate their backup and recovery solutions. After finishing the book I have lots of ideas floating around, but seeing how a one-person, 100-person, 10,000-person, and 500,000-person environment implement BAR would be greatly appreciated.