Backup, part at the request of readers: Overview of UrBackup, BackupPC, AMANDA

This review note continues backup cycle, written at the request of readers, it will focus on UrBackup, BackupPC, as well as AMANDA.

UrBackup overview.

At the request of the participant VGusev2007 adding an overview of UrBackup, a client/server backup system. It allows you to create full and incremental backups, can work with device snapshots (Win only?), and can also create file backups. The client can be either on the same network as the server or connected via the Internet. Change tracking is announced, which allows you to quickly find differences between backups. There is also support for storage deduplication on the server side, which saves space. Network connections are encrypted, there is also a web interface for managing the server. Let's see what she can do:

In the full backup mode, the following results were obtained:

Openning time:

First start
Second launch
Third launch

First test
8m20s
8m19s
8m24s

Second test
8m30s
8m34s
8m20s

Third test
8m10s
8m14s
8m12s

In incremental backup mode:

Openning time:

First start
Second launch
Third launch

First test
8m10s
8m10s
8m12s

Second test
3m50s
4m12s
3m34s

Third test
2m50s
2m35s
2m38s

The size of the repository in both cases was approximately 14 GB, which indicates a working deduplication on the server side. It should also be noted that there is a discrepancy between the time of creating a backup on the server and on the client, which is quite clearly visible from the graphs and is a very pleasant bonus, since the web interface shows the time of the backup process on the server side without taking into account
client state. In general, the graphs for a full and incremental copy are indistinguishable. Probably the only difference is how it is handled on the server side. Also pleased with the low processor load on the redundant system.

BackupPC Overview

At the request of the participant vanzhiganov I add an overview of BackupPC. This software is installed on a backup storage server, written in perl, works on top of various backup tools - primarily rsync, tar. Ssh and smb are used as transport, there is also a cgi-based web interface (deployed on top of apache). The web interface has an extensive list of settings. Of the features - the ability to set the minimum time between backups, as well as the period during which backups will not be created. When choosing a file system for the backup server, you need to be careful about supporting hard links. Thus, the file system for storage cannot be divided into mount points. In general, a rather pleasant impression, let's see what this software is capable of:

In the mode of creating full backups with rsync, the following results were obtained:

First start
Second launch
Third launch

First test
12m25s
12m14s
12m27s

Second test
7m41s
7m44s
7m35s

Third test
10m11s
10m0s
9m54s

If you use full backups and tar:

First start
Second launch
Third launch

First test
12m41s
12m25s
12m45s

Second test
12m35s
12m45s
12m14s

Third test
12m43s
12m25s
12m5s

In the incremental backup mode, tar had to be abandoned, since backups were not created with such settings.

The results of creating incremental backups using rsync are as follows:

First start
Second launch
Third launch

First test
11m55s
11m50s
12m25s

Second test
2m42s
2m50s
2m30s

Third test
6m00s
5m35s
5m30s

In general, rsync has a slight speed advantage, and rsync also works more economically with the network. Some of this can be offset by less cpu usage with tar as the backup program. Another advantage of rsync is that it works with incremental copies. The size of the repository when creating full backups is the same, it is 16 GB, in the case of incremental backups - 14 GB per run, which means that deduplication is working.

AMANDA review

At the request of the participant old add AMANDA tests,

The results of a test run with tar as the archiver and compression enabled are as follows:

First start
Second launch
Third launch

First test
9m5s
8m59s
9m6s

Second test
0m5s
0m5s
0m5s

Third test
2m40s
2m47s
2m45s

The program fully loads one processor core, but due to the limited disk iops on the side of the backup storage server, it cannot develop a high data transfer rate. In general, the setup caused a little more trouble than the rest of the participants, since the author of the program does not use ssh as a transport, but implements a similar key scheme, creating and maintaining a full-fledged CA. It is possible to widely restrict the client and the backup server: for example, if they cannot fully trust each other, then it is possible, as an option, to prohibit the initiation of backup recovery from the server side by setting the value of the corresponding variable to zero in the settings file. It is possible to connect a web interface for management, but in general, a configured system can be fully automated using small bash scripts (or SCM, for example, ansible). There is a somewhat non-trivial system for setting up storage, which, apparently, is associated with support for an extensive list of various storage devices (LTO cassettes, hard drives, etc.). It is also worth noting that of all the programs discussed in this article, AMANDA is the only one that was able to detect a directory rename. The size of the repository in one run was 13 GB.

Announcement

Backup, part 1: Why backup is needed, an overview of methods, technologies
Backup Part 2: Reviewing and testing rsync-based backup tools
Backup Part 3: Review and testing of duplicity, duplicati
Backup Part 4: Reviewing and testing zbackup, restic, borgbackup
Backup Part 5: Testing bacula and veeam backup for linux
Backup Part 6: Comparing Backup Tools
Backup Part 7: Conclusions

Source: habr.com