In the message dated: Sun, 12 May 2013 09:20:18 -0400, The pithy ruminations from "Edward Ned Harvey (lopser)" on <[lopsa-tech] Backup Reliability> were: => Without checking the internet, and before you listen to other peoples' => anecdotes or anything, I'd like to hear your gut feel, I want to know what => your natural instinct is. What do you think about the reliability of the => following tools?
You're asking about the reliability of the tools, but they are probably the most reliable parts of the backup process. Each of the tools you mention has been in development & use for years, if not decades. Any obvious bugs have probably been squashed and a body of knowledge has been accumulated for each tool (which other people have provided in their replies). I'd say that the most important part of reliable backups isn't the tool itself--all the tools you listed are good at copying some 1s and 0s from one device to another, and reasonably good at restoring that data when needed. The most unreliable part of backups is probably the "process", which begins (as you seem to be doing sensibly) with a review of the available tools. However, the reliability and usefulness of any of the tools you list is not intrinsic to the tool, but tightly coupled to the environment--to the use of the tool. For example, while any of those tools can do backups and restores, some are more suitable for small files that change often and where many generations need to be kept, while others may be a better choice for multi-TB backups of almost static data. Once a tool--any tool, not matter how appropriate--is chosen, the most important factor in reliable backups is the process of using that tool. None of the tools will automagically configure themselves, repair errors, change tapes, move external drives to offsite storage without dropping a drive onto the floor, restore data and know what should be overwritten and what should be retained, etc., etc. The daily process of using the tool, combined with the less common tasks of configuring and troubleshooting the tool, are the real keys to reliability. Having ranted a bit, my comments on specific tools are below, for what they're worth... => => => => If you have personal experience with them, obviously, that will shape your => perception. But even with no knowledge or experience with a particular => tool, I still want to know your instinct. Because your preconceived notions => influence decisions you make, give you bias in terms of what tools you even => think about using or researching. I do like that explanation for why this query came about in the first place. => => => => Focus on reliability. ;-) => => => => rsync As a transport mechanism: very reliable. As a backup mechanism--the weakness is in the interface, and the burden on the operator to select option and wrap the tool in order to do backups, particularly multi-generational (ie., more than just a snapshot copy in time). => => rsnapshot => => rdiff-backup => => tar See above for 'rsync', with the added unreliability of historic limitations on path lengths, difficulty in selecting specific files or directories (without using additional tools, ie. "find something | tar"). => => amanda => => bacula I use bacula at $WORK. It's reliability, for better or worse, is directly tied to it's complexity. That means that there's a greater possibility of user error--typically in the configuration, not in the execution of a single command, as with tools like 'rsync'. The least reliable parts of the use of bacula for me are in the interface to the physical environment (media management, addressing drives and the controller in a tape library, etc.). In practical terms, that layer only exists at a particular scale of backups (TB+), and any backup mechanism, whether it's tar or NetBackup, must deal with that complexity. At the same time, the process of dealing with that layer, independant of the tool, is fundamental to the overall reliability. In short, when configured correctly and maintained closely, it's very reliable...but those can be difficult tasks. => => Any other tools that you'd like to mention, that you're likely to use for => backups => => => => Follow-up question: Given that these are all free open source packages, => which are probably included with your "stable" OS distribution, would you => have bias to assume they're reliable, just because of that? Yes, I assume for the most standard use of each tool that they are reliable. Using tools outside their 'core competency' (ie., using tar to backup an interleaved multi-TB datastream from multiple servers") will drastically reduce their perceived reliability....but a poor workman blames his tools...choosing the wrong tool is the first step in a reliable process. => => => => In a follow-up post, momentarily, I'll explain why I'm asking these => question. => Which may be more interesting than the specific question... Mark _______________________________________________ Tech mailing list [email protected] https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
