Results of Amanda Users Survey 2003
This one needs major editing ... sgw
Introduction
AMANDA (Advanced Maryland Automatic Network Disk Archiver) is a backup system freely available available in source form. For more information about AMANDA visit the website http://www.amanda.org.
Readers of the amanda-users mailing list (approximately 1100 subscribers) were asked in April of 2003 to describe their use of the backup system by filling out a survey questionnaire that was posted to the mailing list weekly. Seventy two (72) surveys were submitted during the month-long collection period.
In several postings that follow, I will report summaries of the collected data. In some cases I've included comments (Surveyor's Notes) of my own reaction to the data.
Later, I plan to recast these results in the form of questions and answers to add to the FOM (FAQ-O-Matic) at amanda.org. Should you have any thoughts on this please let me know via the list or personal email.
Jon LaBadie
[Surveyor's Note] While the number of submitted surveys neither surprised me, nor disappointed me, what did greatly disappoint me personally was this fact: Of the twenty five (25) individuals most active on the list (based on number of postings for 12 months), only seven (7) bothered to respond to the survey.
Organizations
What follows is a summarization of the responses describing the organizations using AMANDA.
Types of organizations using AMANDA
Category | Num | Type |
---|---|---|
University Department | 20 | - |
Research Organizations | 10 | |
Internet Service Providers | 5 | - |
Data Processing Services | 11 | software, graphics, integration, consulting |
Manufacturers (DP) | 5 | computers, electronics, semiconductors |
Manufacturers (non-DP) | 4 | rubber, chemical, turbine, videoBoris |
Service Organizations | 4 | publishing, power utility, forestry, money management |
Small Office/Home Office | 6 | - |
Others | 3 | natural history museum, education charity, national opera and ballet |
[Surveyor's Note] The `Categories' were my grouping, they were not on the survey. I expected more in the SOHO category.
[Surveyor's Note] Extent of AMANDA usage was asked on the survey. Few indicated that their entire firm based their backups on AMANDA, and then only relatively small companies (like JG Computing :). Thus the category is `University Department' not `University'. Similarly, AMANDA is used by a groups within organizations like the chemical manufacturer or computer manufacturer or federal government noted above, not the entire entity. But AMANDA may have helped produce the tires you are riding on, the workstation you are typing on, or to collect the taxes you pay :))
Location of organizations using AMANDA
Country | Num |
---|---|
Belguim | 1 |
Japan | 1 |
Poland | 1 |
Canada | 1 |
Mexico | 1 |
Switzerland | 2 |
Estonia | 1 |
Norway | 1 |
Sweden | 3 |
Finland | 1 |
Portugal | 1 |
Germany | 5 |
France | 1 |
United Arab Emirates | 1 |
United Kingdom | 5 |
Italy | 1 |
Netherlands | 2 |
United States | 45 |
[Surveyor's Note] The totals are greater than number of submitted surveys as some respondents described multi-national installations.
[Surveyor's Note] Americans must be procrastinators. For a while, the number of non-US submitted surveys exceeded the ones from the US.
Length of AMANDA usage at organization
Usage in years | Num |
---|---|
0-1 | 25 |
1-2 | 8 |
2-3 | 13 |
3-4 | 10 |
4-5 | 8 |
5-6 | 2 |
7-8 | 2 |
8-9 | 2 |
9+ | 1 |
Summary
10% of responders > 5 years use |
33% of responders > 3 years use |
50% of responders > 2 years use |
33% of responders < 1 year use |
Mean and Median are ~29 months (2.5 years)
[Surveyor's Note] Some really long time users of AMANDA. It must do the job.
Servers and Clients
AMANDA servers
Number of servers in organisation | Num |
---|---|
1 | 53 |
2 | 11 |
3 | 5 |
5 | 1 |
6 | 1 |
7 | 1 |
Operating Systems Running on AMANDA Server Hosts
Operating system (version) | Num |
---|---|
FreeBSD | Total 10, including |
4-STABLE | 2 |
4.7 | 1 |
4x | 1 |
HP-UX | Total 2, including |
10.20 | 1 |
11.00 | 1 |
IRIX | Total 2, including |
6.5 | 1 |
SCO Openserver | Total 1, including |
5.0.6 | 1 |
Solaris | Total 14, including |
8 | 5 |
8 x86 | 1 |
9 | 3 |
Tru64 | Total 4, including |
5.1 | 3 |
Linux | Total 48, including |
Debian | Total 13 |
2.4 | 1 |
3.0 | 2 |
Woody | 3 |
Redhat | Total 14 |
6.2 | 1 |
7.1 | 2 |
7.2 | 1 |
7.3 | 6 |
8.0 | 3 |
8.0 with RHnet kernel | 1 |
Slackware | Total 1 |
7 | 1 |
SusE | Total 3 |
8.1 | 2 |
AMANDA Versions Run on Server Hosts
Amanda Version | Num |
---|---|
2.4.1 | Total 2, including |
2.4.1p1 | 1 |
2.4.2 | Total 33, including |
2.4.2p1 | 1 |
2.4.2p2 | 28 |
2.4.3 | Total 21, including |
2.4.3b2 | 1 |
2.4.3b3 | 1 |
2.4.3b4 | 2 |
2.4.4 | Total 21, including |
2.4.4-20030415 | 2 |
2.4.4-20030426 | 1 |
[Surveyor's Note] The percentage of older versions surprised me. Shouldn't have as I just upgraded 2 months ago. If it works why fix it?
AMANDA CLIENTS
Number of AMANDA Clients
Number of Clients | Host sites |
---|---|
14 | a6-1012 |
25 | b11-2013 |
36 | 21-3012 |
46 | 31-505 |
57 | 61-704 |
Operating Systems Run on AMANDA Clients
(mentions, not number of hosts)
Operating system | num |
---|---|
AIX | total 4, |
Linux (all variants) | total 86 including |
1 | 4 |
Debian | total 17 |
2.2 | 1 |
Darwin | total 1 |
2.4 | 1 |
3.0 | 2 |
FreeBSD | total 17, including |
potato | 1 |
2.2.5 | 1 |
sarge | 2 |
3.2 | 1 |
sid | 1 |
3.3 | 1 |
stable | 1 |
3.4 | 1 |
woody | 2 |
4-STABLE | 2 |
Mandrake | total 1 |
4.1 | 1 |
8 | 1 |
4.7 | 1 |
RedHat | total 26 |
5.0 | 1 |
6.0 | 1 |
6.2 | 2 |
6.2 Alpha | 1 |
6.2 sparc | 2 |
7.1 | 3 |
7.2 | 2 |
7.3 | 4 |
8.0 | 5 |
9 | 1 |
HP-UX | total 8, including |
10 | 1 |
10.20 | 2 |
11 | 3 |
Irix | total 8, including |
6.5 | 3 |
6.5 MIPS | 1 |
Slackware | total 3, including |
7.0 | 2 |
8.1 | 1 |
MacOS X | total 6, including |
10.1 | 1 |
10.2 | 1 |
SuSE | total 4 |
NetBSD | total 1, including Yellowdog |
OpenBSD | total 1 |
Tru64 | total 8, including |
3.2 | 1 |
4.0 Alpha | 1 |
5.1 | 1 |
Snap | total 1 |
Windows | total 22 |
95 | 1 |
98 | 2 |
NT | 4 |
NT 4 server | 1 |
2000 | 7 |
XP | 3 |
Solaris | total 39, including |
2.5 | 2 |
2.6 | 2 |
7 | 5 |
8 | 11 |
9 | 4 |
8 x86 | 3 |
9 x86 | 1 |
SunOS | total 1 including |
4.1.4 | 1 |
AMANDA Versions of Run on Client Hosts
Version | num |
---|---|
2.4.1 | total 4, including |
2.4.1p1 | 3 |
2.4.2 | total 42, including |
2.4.2p1 | 1 |
2.4.2p2 | 33 |
2.4.3 | total 30, including |
2.4.3b2 | 1 |
2.4.3b3 | 1 |
2.4.3b4 | 1 |
2.4.4 | total 25, including |
2.4.4-20030415 | 2 |
2.4.4-20030426 | 1 |
2.4.4b1 | 1 |
Installation Properties
Total Disk Capacity of Clients
smallest: 20 GB
largest: 5000 GB
Capacity | num |
---|---|
20 to 100 GB | 13 |
101 to 200 GB | 22 |
201 to 500 GB | 16 |
501 to 1000 GB | 12 |
1001 to 2000 GB | 5 |
2001 to 5000 GB | 6 |
Total Data Currently Stored on Clients Disks
smallest: 3 GB
largest: 4000 GB
Capacity | num |
---|---|
3 to 50 GB | 22 |
51 to 100 GB | 14 |
101 to 200 GB | 15 |
201 to 500 GB | 10 |
501 to 1000 GB | 7 |
1001 to 4000 GB | 5 |
Average Size of a Single Backup
smallest: 0.3 GB
largest: 700 GB
22 0.3 to 10 GB
11 11 to 20 GB
22 21 to 50 GB
13 51 to 100 GB
8 101 to 200 GB
1 350 GB
1 700 GB
Backup Program Used
72 use gnutar or some tar variant including:
3 hfstar
1 nttar
1 smbtar
50 use dump or some dump variant including:
1 e2fsdump
2 ext2dump
12 ufsdump
2 vdump
1 vfsdump
2 vxdump
5 xfsdump
Dumpcycle
5 0 days
3 1 days
1 2 days
6 3 days
5 5 days
34 7 days
17 8 days - 2 weeks
10 15 days - 4 weeks
5 30 days - 75 days
1 4 months
1 6 months
1 1 year
1 1000 days
Frequency of Backups
6 1/week
4 2/week
4 3/week
5 4/week
36 5/week
6 6/week
31 7/week
2 1/month
Type of Compression Used
1 gpg
15 none
17 hw
31 sw - server
46 sw - client
Installed Hardware
Brand and Model of Tape Drive Used at AMANDA Installations
The number of mentions of each brand is shown
followed by specific models where reported.
3 Archive, including: 2 Qualstar TLS-4200
Python 04106
Python 06408 10 Quantum, including:
DLT-VS80
5 Compaq, including: DLT-4000
AIT-2 DLT-7000
DLT-90 DLT-8000
MSL5126-S2
SDLT320 10 Seagate, including:
DAT
1 Dell DLT-IV 12/24G
20/40G
5 Ecrix, including: CTL-96/4586N
VXA-1 DDS4
DDS3 Scorpion-24/STD124000N
5 Exabyte, including: 13 Sony, including:
8500 AIT-1
Eliant 820 AIT-2
VXA-2 AIT-3
DLT-IV
15 HP, including: DLT-9000
A4845A SDX-500C
C1533A Sony SDX-700C
C1557
C6280-7000 3 Sun, including:
DDS-4 DDS3
DLT vs80 7000
ESL9595 Python
SureStor DAT-24
SureStor DAT-8 7 Tandberg, including:
Surestore Ultrium 230e DLT vs80
Ultrium 1 SDLT 220
SDLT-320
1 IBM DDS4 SLR5
SLR7
1 Mammoth
2 Wang, including:
3 None, backup to disk only DAT 3400DX
2 Overland Data, including:
Loaderexpress DLT8000
Native Tape Capacity of the Above Drives
7 4 GB 8 35 GB
1 7 GB 13 40 GB
2 10 GB 8 50 GB
11 12 GB 1 60 GB
1 15 GB 1 80 GB
1 17 GB 7 100 GB
10 20 GB 1 110 GB
1 22 GB 3 160 GB
4 25 GB 1 500 GB
2 30 GB
Preferred Brand of Tape
2 Compaq
1 Dell
4 Exabyte
3 Fujifilm
5 HP
1 Imation
2 Maxell
2 Quantum
11 Sony
1 Verbatim
16 No Preference
Tape Format
1 4mm 1 EXA
4 8mm 3 LTO
2 AIT 1 Mammoth
1 AIT-1 4 sDLT
7 AIT-2 1 sDLT320
3 AIT-3 1 SLR
8 DDS-2 1 SLR25
12 DDS-3 1 SLR7
6 DDS-4 1 Travan-20
15 DLT 1 VXA
1 DLT-80 1 VXA V-23
7 DLT-IV 1 VXA-2
Tape Changers/Libraries Used at AMANDA Installations
[Surveyor's Note] A total of 28 of 72 respondents reported having using tape changer or library. Several of those reported having more than one. Surprisingly, the 28 respondents reported 26 distinct models of changer/library.
The table below shows the models reported by 28 respondents. Where multiple reports for the same model existed, only one entry is show. Where the multiple reports showed different configurations (eg. 1 drive vs 2 drive), only the larger configuration is shown.
BRAND MODEL # of # of total
drives tapes capacity
ADIC ?? 1 4 200 GB
ADIC FastStor DLT 1 7 245 GB
Compaq MSL5126 1 25 4000 GB
Compaq SSL2020 2 20 1000 GB
Dell Powervault 122T 1 8 320 GB
Hewlett-Packard A4845A 5 588 20000 GB
Hewlett-Packard C1557A 1 6 72 GB
Hewlett-Packard ESL9595 2 595 70000 GB
Hewlett-Packard MSL5026S2 2 26 4160 GB
Hewlett-Packard SureStore 6x40 1 6 120 GB
Hewlett-Packard SureStore 6x24 1 6 72 GB
Overland Data LXB 4120 1 10 150 GB
Overland Data Library Pro 1 19 1900 GB
Overland Data LoaderXpress 1 15 525 GB
Qualstar 6220 2 20 700 GB
Qualstar TLS-4200 2 20 2000 GB
Quantum DLT7000 2560 1 8 320 GB
Segate CTL-96/4586N 1 4 14 GB
Sony TSL-9000 1 8 96 GB
SpectraLogic Bullfrog 10000 3 40 10000 GB
SpectraLogic Treefrog 2000 1 15 750 GB
StorageTek Timberwolf 9730 2 30 1000 GB
Straightline Harrier/830 2 30 1500 GB
Sun L20 2 20 2000 GB
Sun L40 2 40 1600 GB
Sun Storedge L9 1 9 360 GB
Special Feature Usage
What follows is a summarization of the responses describing the
use of two special features of recent versions of AMANDA, the
"file:driver" to use a hard disk drive as a backup device and
RAIT (Redundant Array of 'Inexpensive' Tapedrives) to allow
mirroring or striping of backups across multiple drives.
The file:driver
Do you use the <file:driver>?
53 no
11 yes
5 what is the <file:driver>
If not, do you expect to?
32 no
14 yes
8 maybe
RAIT
Do you use RAIT?
68 no
0 yes
If not, do you expect to?
40 no
8 yes
3 maybe
[Surveyor's Note] I was mildly surprised, and amused, by the 5 "what is the <file:driver> responses. And considering those responses, but the lack of similar responses to the RAIT question.
[Surveyor's Note] No respondent uses RAIT! Disappointing. I wonder if any site is currently using this feature.
Non UNIX Clients
What follows is a summarization of the responses describing the
various methods to backup 'non-unix' clients, mostly PC's running
versions of Windows.
PC's Backed Up Using SAMBA Connections to AMANDA Clients
A total of 22 survey respondents (of 72) report
they use amanda and samba to backup PC clients.
10 have 1 PC client
10 have 2 - 5 PC clients
1 has 15 PC clients
1 has 40 PC clients
[Surveyor's Note] From these results it seems to me that amanda/samba
is seldom a first choice for backing up PC's. Instead it seems to be
used when a site has an existing amanda installation and would like
to implement some form of PC backup.
The respondents mentioned the following Operating Systems
were being run on the PC clients. The number is the number
of mentions, not number of PC's.
1 Windows 95 2 Windows 2000 Server
1 Windows 98 5 Windows XP
3 Windows NT 1 Windows XP Pro
2 Windows NT 4 1 Snap Server
9 Windows 2000
[Surveyor's Note] Of interest to users running Lotus Domino: One respondent reported using "The Open File Manager" from http://www.stbernard.com to create online backups of their Lotus Domino database files which then could be backed up to their amanda host using samba.
PC's Backed Up Using Cygwin/AMANDA Client Software
A total of 4 survey respondents (of 72) report
they use a cygwin/amanda client to backup PC clients.
2 have 1 PC client
1 has 3 PC clients
1 has 15 PC clients
The only PC operating systems mentioned were
Windows 2000 and Windows 2000 Server.
PC's Backed Up by AMANDA Using Other Techniques
A variety of techniques, other than samba or cygwin,
were described by respondents to backup PC's
1. Two respondents installed an NFS server on Windows 2000
and then NFS mounted the PC disks on the amanda host
and backed them up as local file systems.
2. Similarly, two respondents mounted PC shares of Windows 95,
Windows NT, and Windows 2000 on the amanda host using 'smbmount'
and backed them up as local 'smbfs' file systems.
3. Several respondents reported they first backed up their PC's
(Windows 98, Windows 2000, Windows XP, Windows 2000 Server)
to disk files on the PC's and then backed that file up using
amanda; presumably using samba or some file transfer technique.
One mentioned that the backup stream was sent directly to a
Linux host and backed up there as a local file.
The backup programs the respondents mentioned included:
'Windows backup program'
'BackupPC', a Linux opensource program
'Win2000 Backup Tool'
'Backup Exec'
4. One respondent noted their firewall was a PC running linux.
That system was mirrored using rsync and the mirror was backed up.
Other Backup Techniques Mentioned
Five respondents described backing up systems using "non-traditional"
methods. These methods were used to backup systems running either
MacOS or Windows. They included:
1. MacOS 9, mounted (NFS?) onto a Linux host and backing up the mounts as
local filesystems.
2. Backing up 37 MacOS 8 and MacOS 9 systems using the program Retrospect.
The output of Retrospect was transferred to a Linux host using ftp and
preserved with amanda.
3. Build a native amanda client on MacOS X and use hfstar instead of gnutar.
4. Backup 6 Windows XP hosts using a native windows amanda client.
5. Backup 12 Windows 2000 Server hosts using a native windows amanda client.
This respondent also noted their intention to switch to a cygwin/amanda
client in the near future.