Bacula vs Amanda?

[mlewis]

Anyone here using bacula over or with amanda on their networks?

Just wondering which is the least harder to learn and put to use.

Mike

[DJ]

Well, this is the Amanda forum, so you might get a biased response here, but let me point you to a few facts:

- Usability and ease of installation of Amanda. Here is the link to a paper that provides specific instructions how you can start backup of 2 different Linux and 1 Windows machines with Amanda in less than 15 minutes: [url]http://www.zmanda.com/quick-backup-setup.html[/url]

- The browser based management console in Enterprise version of Amanda makes the installation, configuration and management of backups and recovers very easy.

- If you want to learn more how Amanda compares to other backup products register for free to [url]http://network.zmanda.com[/url] and read the white paper "Advantages of Amanda over Proprietary Backup". It will help you understand some key Amanda differences, and will help you evaluate and transition to Amanda.

Regards
Dmitri Joukovski

[squadra]

Bacula uses MySQL oder PostgreSQL as Backend which can be really ugly... Bacula SQL Kaboom

[mlewis]

I agree that I might get a biased view but I'm not asking someone who is biased, I'm asking if there is someone here who has/is using both.

My problem is that I've started out with bacula and for the amount of time I have, it seems to have a vert steep learning curve. I've gotten the software to work, I've gotten it to see my tape device and work with it but now I need to learn about using it to actually back up.

Before I invest yet more time into it, I wanted some thoughts on this project, non biased if possible since we're all adults .

Mike

[mlewis]

Bacula uses MySQL oder PostgreSQL as Backend which can be really ugly... Bacula SQL Kaboom

I think that's a biased view based on your own experiences. I've been on their mailing list for months now and have not even seen one message about mysql problems.

I'm not asking to trash either package, I'm just wanting to learn about them. Think about it this way, we all want open source to work so that it's not the man who owns it all. It's difficult to find out what those choices are without input from users in various camps.

Mike

[choogendyk]

That's a difficult question (the unbiased part), because we all have our preferences and reasons for choosing the way we have. If you want to see my biases (although I think they are well considered biases ), see:

[URL="http://blogs.umass.edu/choogend"]Ten Things I Like about Amanda[/URL].

I've been on the Amanda list for about 2 years and on the Bacula list for over a year. I've contributed to both. When I finally got the budget for a tape changer late in 2006, I chose to go with Amanda. I had been leaning that way already, but I had been asked to evaluate the alternatives. I had also used Retrospect for many years in a Mac software company, I have written extensive scripts to run my own automated backups in Solaris, and I've had some exposure to Veritas NetBackup.

Explaining why I chose Amanda over Bacula could fill a lot of space. It's really the overall feel of the program and the community. Amanda is focused on the task at hand (give me my backups, please) rather than getting tangled up in all the side issues, such as duplicating native tools by writing its own, developing GUI's, interfacing with SQL databases, etc. These side issues consume developer time, introduce additional classes of errors and bugs, add load to the backup server, and are additional things I would have to deal with as a user.

Of particular note is Amanda's algorithm for planning backups. If peak load in terms of network bandwidth, time to completion, tape use, server load, etc. are issues for your backups, then Amanda's planner is the solution. Sysadmins who don't use Amanda spend significant amounts of time planning and tuning backups (unless they have a real excess of resources), worrying about full backups running too long, a partition filling up here and not getting used there, etc. Amanda juggles all that for you. It smooths network usage, tape usage, and server load over the dump cycle (say, 7 days). If a partition fills up, it will shift things around to maintain the balance. For example, instead of using 3 tapes on the weekend for fulls, and then 1/4 of a tape each week day, Amanda would smooth that out and end up using, say, 3/4 of a tape a day. You'd end up using them more efficiently and saving 2 tapes a week. No sysadmin has the time to sit around figuring that out and adjusting it manually week after week. For me, that's the most magical and satisfying thing about Amanda.

Of course, reliability and minimum exposure to errors or bugs is paramount for any backup software. This is a significant reason why relying on native tools (dump and tar) is important. You have a much better chance of being covered without errors or bugs on whatever platform you use, whether it be Solaris with ufsdump, Linux with gtar, AIX or whatever. While I think Bacula has done a good job, given their choices, I have seen major bugs and errors acknowledged on their list during this year that arise from data not being recorded correctly in the SQL database and from ACLs not being picked up correctly. These are classes of errors that Amanda does not generate. I've also seen a fair bit of energy expended on performance and tuning issues related to the SQL database in Bacula, with that sometimes being the bottleneck for backups. Some of those issues may be resolved in the newest releases, but, nevertheless, they are issues a sysadmin has to be concerned with.

The Amanda community has grown stronger since Zmanda was founded. Rather than being dependent on one core developer, Amanda now has a team of programmers working on development and contributing code back to the community. That, combined with the inherent strengths of Amanda's concept and design, gives me a lot of faith in Amanda's future.

Of course, all of this is from my perspective and with respect to the environment of servers that I manage. I manage mostly Solaris servers with some additional mix of Unix and BSD. At the moment, I'm not covering any Linux servers (although that would be prime territory for Amanda), and I don't do Windows. My desktop is Mac OS X.

I hope that helps.

[mlewis]

Wonderful, you've given me a lot to consider and think about.

My application requires backing up pretty much only Linux machines, RHEL and CentOS servers. While there are a few Win machines, they run their own backups and those files could be stored on the Linux servers which are being backed up.

I don't have much choice but to give this a lot of thought considering the complexities of backing up a complicated network. I've spent a couple of years building a network of clustered web servers, high availability networking, the list is too long to type. I've had to learn every single piece of hardware and software that I've put together as this is a private project.

The key has been finding software which I can learn quickly, not having to become a total complete master of each solution which I use. Often when one asks for help, they are told to RTFM or they are told that if they aren't willing to take the time to learn the software, why ask for help? That seems like the wrong answer to me if we're all supporting open source. One does not have time to become a pro at everything that they touch, there is simply too much which makes up a complex network.

Anyhow, I'll take a closer look and ask more questions but I'm in a catch 22 since I have little time and yet must start looking at backups seriously.

I was able to get bacula working with my 380 tape Qualstar but when it came time to learn about backing up, I ran out of time. I'm hoping that amanda might be a little less of a learning curve.

Thanks again for all of this input, it's very much appreciated.

Mike

[choogendyk]

What you are suggesting for your Win machines is actually what I do for the labs in my department. They have more Macs than PCs, but I give them responsibility for backing up to drives on my servers. They buy a drive and some tapes, I set it up, they run their backups, and then I periodically archive them to tape with Amanda.

The work you've done so far will fit right into Amanda. Getting tapes to work for your OS is the same for both Bacula and Amanda. In both cases, once you can use a tape with mt, you are set. If you have a tape library, once you can use it with mtx, you are set. There is a glue script chg-zd-mtx that will connect Amanda to mtx. You configure that to point to the library robot device. Then in your amanda.conf, you point to the changer script and to the tape device.

If you don't have much time, take it a step at a time. Set up the backup server to just backup itself. Then figure out how to add a client. Then add another client. That also gives Amanda time to smooth out the full backups, since it isn't starting them all at once. Give yourself a good size holding disk on the backup server so that all the parallel incoming backups can be spooled efficiently and the tape can be written at maximum speed.

And, be sure to holler for help if you need it.

[mlewis]

>you can use a tape with mt, you are set. If you have a tape library, once >you can use it with mtx, you are set. There is a glue script chg-zd-mtx >that will connect Amanda to mtx. You configure that to point to the library >robot device. Then in your amanda.conf, you point to the changer script >and to the tape device.

I quickly messed around with the 'get it running in 15 minutes' but got nowhere . And yes, I do have it working with bacula so sounds like a good start.

Is there another document you can suggest which has only the server and one Linux client perhaps?

Mike

[choogendyk]

You'll find all the user documentation resources here:

[url]http://wiki.zmanda.com/index.php/User_documentation[/url]

What I used when I was getting started was the quickstart, which is the second link down on that page and is here:

[url]http://wiki.zmanda.com/index.php/Quick_start[/url]

As I recall, I had already read enough stuff to be familiar with the overall issues and configuration, and I made a few forays into the rest of the documentation as I was following that Quick_start. But the Quick_start was the basic guide that I followed to get things going.