Any Slackware Server Stats?
 

Gonna start off really simple (isn't that the Slackware way?) and just ask it outright since I can't seem to find anythong definitive on the subject, and I'm aware it can be a nebulous subject that presents difficulty in gathering evidence. This is probably especially true since RHEL apparently dominates. Nevertheless, I'd like to have something more than my own prejudices at work so....

Does anybody know, or know where I can find, statistics on the relative "market share" of Slackware servers? for example as compared to Debian?

A good source for this is on http://distrowatch.com/

They currently have Slackware listed in 14th place. I'm not exactly sure how they collect their data, but it's a good start.

from distrowatch.com:

Rank Distribution H.P.D*
1 Ubuntu 2022>
2 Mint 1517>
3 Fedora 1443>
4 openSUSE 1204=
5 Debian 1083>
6 Sabayon 787>
7 PCLinuxOS 776<
8 Arch 734>
9 Puppy 719>
10 Mandriva 657<
11 Ultimate 566>
12 CentOS 549>
13 Lubuntu 544<
14 Slackware 495>
15 FreeBSD 466>
16 Tiny Core 451<
17 Chakra 438>
18 MEPIS 434>
19 Peppermint 408>
20 CrunchBang 379>
21 Gentoo 374>
22 Red Hat 346>
23 Vector 342=
24 Kubuntu 324=
25 Salix 312=
26 Super OS 309<
27 PC-BSD 296>
28 aptosid 294=
29 KNOPPIX 292>
30 Zenwalk 281>
31 Dreamlinux 274=
32 MeeGo 269=
33 BackTrack 265=
34 Unity 261<
35 Zorin 247>
36 Xubuntu 241=
37 Ubuntu Studio 228>
38 Frugalware 224>
39 Parted Magic 223>
40 Pardus 222>
41 gOS 219=
42 ClearOS 208<
43 ArchBang 206>
44 Zentyal 202=
45 ZevenOS 197=
46 Clonezilla 189=
47 SliTaz 182>
48 Scientific 182>
49 Macpup 176=
50 moonOS 175>
51 OpenBSD 170=
52 Absolute 166=
53 SystemRescue 160=
54 TinyMe 155<
55 Quirky 154=
56 wattOS 150>
57 Mythbuntu 150=
58 GhostBSD 146<
59 Slax 144=
60 EasyPeasy 141=
61 DragonFly 141>
62 PC/OS 140<
63 Pinguy 139>
64 MINIX 139=
65 LinuxConsole 139=
66 64 Studio 136=
67 Calculate 135=
68 FreeNAS 130=
69 Lunar 129<
70 Alpine 129>
71 Element 126=
72 Deepin 124=
73 Solaris 121=
74 VortexBox 120=
75 Elive 120=
76 Jolicloud 116=
77 AUSTRUMI 116>
78 RIPLinuX 115=
79 LFS 115<
80 GParted 114>
81 Yellow Dog 112=
82 linuX-gamers 111=
83 NetBSD 110=
84 Imagineos 109=
85 Untangle 108=
86 Parsix 108=
87 NuTyX 108<
88 ALT 108=
89 Nexenta 103=
90 Novell SLE 102=
91 DEFT 102=
92 Ubuntu Christian 101=
93 PureOS 101=
94 Linux XP 101>
95 Connochaet 101=
96 Turbolinux 100>
97 Moblin 100=
98 Linpus 100=
99 BlankOn 99=
100 Kongoni 95<
*H.P.D = hits per day
Updated daily
More statistics...

Thanks MrBob...
 

(For the try)

That list is page hit ranking which means any script kiddie in the world can bump it up because he heard "Backtrack4 is the Bomb" and so simply by popularity the list is heavily biased toward desktop systems, not serious server implementation. For example Puppy is a great little LiveCD of around 200MB that is trying to transition from live environment only (and may succeed since it is decent on really old laptops) but is by no means suitable for server work. Technically any Linux can be added to in order to handle server work, but there is just no need since there are distros designed for that market, or like Slackware, a solid core for just about anything.

Distrowatch stats have nothing to do with a server market share. They base their statistics on the number of hits on a website.
I don't think it'll be possible to get even a rough idea about the distro breakdown for linux servers. After all, I guess only Red Hat / Novel customers can buy support (something to lean on when counting servers). What about servers running debian, CentOS, Slackware where no financial transaction is made and therefore, nothing to lean on when compiling data. You may find some data on netcraft.com but it'll probably put every linux distro in one bag.

Thanks dr.B-O-B :)
 

That statement was completely subjective and debatable. In fact, I tend to see CentOS deployed more along the numbers we see in that hit-meter above, where RH comes in at #22.

I also see less deployments of subscription based support distros than free distros in general.

And I also see a proportionate level of rise in the lack of competencies when these subscription based distros are incorporated into the enterprise - but hey, after all, they are paying the vendor for their expert support ;)

When I ask, "what various distros or UNIX flavors are incorporated here in this infrastructure?"

When I get a replay back like, Debian, Arch, Slackware, and a few others, I know that there is generally a higher level of competence for the average admin at this company than there is at a comparable site based on RHEL contracts.

um... BackTrack 4 is da bomb, actually ;)

The closest thing we've ever been able to glean the sort of data you're asking for (to my knowledge), is the Linux Counter, and perhaps The Gartner Group.

Hope that helps

A Con Census?
 

I suppose the first clues were "probably" and "apparently" ;) While it is obviously impossible or at least highly unlikely that there is any means of discovering how many Linux desktops are in regular use, I imagine since all manner of other stats are kept on businesses which, by definition, are more public and higher profile, and for whom such data might be valuable for gathering capital - services, personnel and ultimately income, that at least an estimation is possible based on related objective data. Aren't there numbers for at least how many RHEL certs have been awarded?

The CentOS wiki claims that

"CentOS is the most popular Linux distribution for web servers with almost 30% of all Linux servers using it"

and hopefully such statements aren't merely pulled out of an orifice, in which case somebody is gathering data
IBM for example claims

"Linux brings open innovation to all IBM server and storage system platforms, freeing datacenters from vendor lock-in with choice and flexibility to scale your business on the fastest growing operating system in the world.". Source

Surely they base that on something other than wishful thinking or conjecture. Where did they get their data? That's what I hope to see.


Now THAT is interesting and useful... limited to one perspective, but apparently informed and at least a actual sampling, even if undocumented.

I'd say that is only logical since I expect people and businesses who always take their cars to mechanics have chosen to vacate that area of expertise and "job it out to pros".

I wasn't trying to imply that Backtrack4 is somehow less than serious or valid, only that it attracts every kid looking for a shortcut to some ersatz Angelina Jolie. ;) I built my own years ago on a box still running Slackware v10.2 upon which I gathered the tools I needed such as Nessus (which used to be free), Snort, Tripwire, and Rocket, and others gleaned from Knoppix and other forensics niche distros, so I'm probably biased. The point was that Distrowatch's list is only a sort of popularity contest, heavy on Pop, as in "current buzz", not actual deployment anywhere for anything let alone as servers.


Gartner is interesting if obtuse on Linux, but Linux Counter only serves to highlight the problem since they haven't been updated since 2005. Similarly Distrowatch has very old information reported as authoritative some even bordering on FUD. Case in point

It does some. At least it could be the start of some dialogue in this thread. Thank you.

Not always true. In larger businesses/corporations the ground-level admins will have to conform to company wide standards, and will be forced to use stuff they wouldn't normally touch with a 10 foot Battle-lance. You can't decide competence based on something like this. Having said that, there's a hell of a lot of incompetence out there thanks to the "Pay peanuts, get monkeys" factor.

The Top 10 Linux Server Distributions — ServerWatch.com
http://www.serverwatch.com/columns/a...tributions.htm

More of an opinion piece article.

Websites using Linux, but is that the same as server usage?:

http://w3techs.com/technologies/deta...-linux/all/all

And I can't see those listed below SuSE as being more suitable than Slack for a server.

Yikes!
 

Well that site is definitely an eye-widener! I agree that the list is not "seemly" but the page on how they gather data is fairly reasonable. This deserves more research. Thank you!

Several of the folks who left comments there just about had a kniption [sic] fit about Ubuntu coming out on the top of that list, questioning the validity of the findings based upon their own professional experience and observations.

I agree with those notions. Ubuntu is, IMO, a n00b distro (and I mean that w/no disrespect other than I believe people should run Debian and not Debuntu, as I say).

Ubuntu is great for the n00b - bells whistles, easy install, protection (FWIW) against someone running as root, etc. - just what the mACROsFOT point-n-clickers need to get off the ground.

In short, Ubuntu is the new "Mandrake".

Back when everyone had rpm fever (before rpm hell), There was Mandrake-Linux, which was what I used to recommend for first time UNIX converts to set up as their desktop environments.

Ah, but then they went public, did the dot-com faceplant, and yet Mandriva, the Mandrake phoenix, appears to have a great share of the market to this day.

Yup. I'm a Slacker, but Debian is used in shops needing high availability that are run by open-source zeaolots of the kind that cringe even at some of the packages in Slackware (i.e., pine/pico vs nano and firefox vs iceweasel).

I also just read some stats about IIS vs Apache, and Internet Exploder vs real browsers, and the only point really significant to this thread on that point, is that CentOS really leads the way with the lionshare of the commercial hosting industry.

hm....

Okay that might be for a couple of reasons that I'll address:

• CentOS isn't RHEL - (meaning, it's community, not subscription based).

• CentOS, by virtue of being a non-subscription based Linux distro, is "chosen" by competent admins, as a matter of choice, rather than a matter of corporate standards (as alluded to in a post above)

hm... so why isn't Slackware (scratch that, we know why) or Debian or Arch listed as being so prominent in the commercial hosting industry along with CentOS - and remember, RHEL doesn't even show up as a blip on this radar screen.........

Answer:

• cPanel/WHM licensing

Why would someone pay for Redhat, then pay for cPanel (and then even Parallels if their selling VPS solutions?)

Again, and I acknowledge that it's not "always true", as pointed out above, but competency levels come right back into the discussion when the commercial hosting industry is included in the tally of the most prevalent distros.

BTW: Linode has Slack images for their VPS's, even though it isn't -current or -13.1

To summarize, when an rpm distro needs to be chosen for commercial deployment by people whose level of competence rises above that of a closed vendor solution - CentOS reigns supreme (So when the recruiter asks you if you 'know' Redhat, you say, "Sure, but I USE CentOS").

When a seasoned IT manager has the discretion to standardize or choose the distros of his choice in the enterprise, she chooses:

• Debian GNU/Linux (if she's a GPL zealot)
• Other great distros, maybe several different ones including other Unices such as BSDs and Solaris. Slack/390, etc.

When an uninformed bureaucratic committee decides upon the standardization in an enterprise, you end up with RedHat, Oracle, SuSE, and other distros which, although good, come with support strings attached.

Actually, I've always had a warm spot in my heart for SuSE - considering it's simply a German fork of Slackware (historically, anyway). And it might be the one thing that Novell has done well in the last ten years too.

I wouldn't hesitate to recommend Ubuntu to a n00b who I felt was really interested in migrating to a Linux Desktop (although I would still prolly say Debian), and when people ask which is the best for a n00b, I (to the astonishment of many) say Slackware.

Great post, very informative. Do you know, by any chance, is there like a somewhat complete list of Slackware packages zealots cringe at? I was musing for a while about making such a list, but I am very lazy and I don't want to start from scratch.

Has nothing to do with OP, interesting? Maybe not.........?

http://www.google.com/insights/searc...e=&clp=&cmpt=q

A very interesting question has been raised...
 

That might actually be best addressed as a new thread. I'm sure that many would love to weigh in on that, however, it should also be noted that the *contrib* label can be manually enabled for apt to find those things we're going to use anyway ;)

No!
 

That's just downright depressing :( a certain sign of de-evolution if there ever was one