Slashdot Mirror
Apache Now Runs On Over 5 Million Sites
According to the December Netcraft survey, Apache can now be found running on over 5 million sites. Overall, Apache's "market share" dropped about a third of a percent, with the biggest change being a 0.77% increase by Zeus mostly due to its use by UUNet.
oops, should have titled that "what e-commerce sites are really running".
heheh.
Anyways, plenty of startups are running Apache; it's really only the stuffed-shirt suits that persist in trying to deploy on IIS. (ick, poo)
Remember that what's inside of you doesn't matter because nobody can see it.
Amazon.com:
Server: Stronghold/2.4.2 Apache/1.3.6 C2NetEU/2412 (Unix)
CDNOW.com:
Server: Apache/1.3b5
Xoom.com:
Server: Apache/1.3.6 (Unix) secured_by_Raven/1.4.0
MP3.com:
Server: Apache/1.3.3 (Unix) mod_oas/4.61
www.jcrew.com:
Server: Netscape-Enterprise/3.6 SP2
www.etoys.com:
Etoys web server 1.2 (hacked version of apache)
www.ibm.com:
Websphere (hacked version of apache)
Who's next?
Remember that what's inside of you doesn't matter because nobody can see it.
Ask him if you should then back Opera over MSIE and Netscape Navigator, since it is the most sold (rather than downloaded for free) browser, and thus by his logic the "market leader".
While the success of Open Source software in the commercial arena is a valid benchmark of the strength of the Open Source software movement, please don't make the mistake of thinking that success in the ecommerce arena is the ONLY valid area of competition.
I do NOT use the web only to buy things. In fact, about 99% of my use of the web is NOT for acquisitory purposes, but for informational ones. I don't care if Microsoft owns (now or in the future) all the ecommerce mindshare, because selling things is NOT what the internet is about! My DSL connection gives me a static IP address, and a consistently accessible platform from which I can publicize my thoughts. The Linux OS gives me an environment that I can take for granted, so that I can concentrate on the message, not the medium. The Apache web server gives me the means to say whatever I want to say, in exactly the way that I want.
The success or failure of Open Source software has NOTHING to do with the success or failure of commercial companies. It has everything to do with your ability to use that software in the way you want, to say what you want. Even if every ecommerce site on the web were running Microsoft's latest version of IIS, the Apache web server would still be a resounding success, because it would still permit me, Joe Average American, to publish my opinion, to use the Internet to communicate to anyone who will listen.
those "piss-ant" sites, those "Here is a Picture of My Cat" web sites, are much more important than you might think. Boring or unimaginative though they may be (I'm a firm believer in the theory that 90% of everything is crap), those piss-ant sites are the mundane embodiment of the real potential of the internet - the ability of the average person to express an opinion, in a non-instrusive way. If MS's products let you buy things, but Apache lets you say what's on your mind, then Apache is a winner. If the Apache web server, in conjunction with the Linux OS, lets you say whatever you want, no matter who owns what percentage of the market... well, there's simply nothing to complain about.
It doesn't matter what you own,
it matters what you do with what you own
- Mike McCafferty
So, if you took the time to notice the small blurb below the fancy graphic:
So to be included you merely have to test a site you are curious about at some point.
What this says to me is that I can put in any number of IP addresses that are bound to a hosting server. If they are doing checks on this alone they would get different points on the graph. However, the problem with that is that a massive virtual hosting operation could really skew the number.
Netcraft even realizes this and states it clearly on the Mechanics page.
Now, the same is true of IIS4 which just makes me think that the number isn't completely representative and that the sample has potential for being flawed.
There are several caveats when looking at compiled data like this in a simplified graphic.
So, before you shoot off at the mouth to your MIS director make sure you can qualify the data you present. Make no mistake that MS has some idea of how many NT4 installs took place with IIS4 counting as running as a service whether or not the people running it know about it. There are many many places with the MS equivilent of the Apache successful install home page. *grin*
Just remember that joe business owners internal intranet IIS4 box with some canned application isn't going to show up on this kind of study. And internal MIS projects that go bump in the night won't either. How many users would really notice if you mapped let .asp be a known extension for perl cgi's? *grin*
http://www.mp3.com/fudge/
http://fudge.org
We've all seen the stats on the number of hosts running PHP on top of Apache, since it's often mentioned in the server header on Apache.
What I'm more interested in is a thorough breakdown of how many sites are running ASP, PHP, Cold Fusion, CGI, FastCGI, JSP/servlet engines and Java app servers.
This can be hairy-to-impossible to measure; all you can really go on are substrings in URIs, and there's no way to tell if something called "foo.pl" is a Perl CGI or a mod_perl module.
What makes this a point of interest is that what HTTP server you're running is an ever smaller piece of the story these days. After all, you can run PHP on IIS. You can run ASPs on Apache under Unix--even VBScript ones, if you buy Chilisoft's module. Servlets, JSPs and high-end application servers are cross-platform: you can most servlets, or something like StoryServer, WebLogic or Dynamo on IIS, Apache or a Netscape server on any of several operating systems without changing a line of your code.
Heck, all those ".cfm" URLs are pretty ambiguous these days. ColdFusion, though closely associated with NT environments, has run on Solaris for ages, the core engine is also now available as a platform-independent servlet, and starting with a beta of 4.5, the flagship product is now also on Linux.
The important NT "wins" are the sites running ASP on IIS, with the logic tier built as COM objects hosted on MTS. And that's how many larger IIS sites are built these days. And unless, like Barnes and Noble, you are getting direct development assistance from Microsoft, this combination should strike a sensible technologist as an alarming degree of lock-in to a single vendor.
Compare this to the freedom that using open technologies, whether PHP and mod_perl, or servlets and JSP, or even large-scale Java app servers gives you. Now that the major app server vendors have coalesced around the J2EE spec, you can switch from one big, scary app server to another, change operating systems and development tools, change databases, and change HTTP servers and still keep the bulk of your code intact. And most of what you're running will also run on a free, dinky little servlet engine or one of the forthccoming open-source EJB app servers chugging down the pike.
Doing things with ASP and don't like the latest direction MS Visual Studio is taking? Don't like the wholesale changes to the VB object model every 3 years? Want to try a different OS because you just can't get decent uptime and clustering working well enough under NT or Win2000? Too bad.
Where IIS is execeling is servers that are actually used by many (read: generate a substancial profit) choose IIS.
I'm not sure that's entirely true. Apache is used by sites such as Amazon, Etoys, and the like, and it would be hard to argue that these sites are marginal, or don't generate substantial profits.
Nunc Tutus Exitus Computarus.
Netscape!
;)
:)
At least, when I worked for the IBM e-business division, all of the major sites the used a unix web server pretty much used either Netscape or Domino. Very few used apache. Netscape is threaded, ie designed to work well on large, multi-processor boxes.
I would say only about 15% of the major IBM customers used NT on the front end, and MAN, what a mistake it was for them, across the board.
This is not, btw, an endorsement of any particular server - just the facts as I remember em.
The web server in an e-commerce site, for those of you who haven't implemented one, is almost an afterthought. Much more important are things like the database, switching/routing, load balancing, and the boxes on which the w/s runs.
Most of the really large sites have 3 or 4 web servers with either a cisco local director or IBM's e-net dispatcher in front of em.
--
Blue, formerly root@justabouteverybigecommercesiteinamerica.com
i browse at -1 because they're funnier than you are.
This is because it shows, when a company's profitability is on the line (and make no mistake, the main purpose of businesses is to make money), who they trust most to make them money (which includes doing it in a safe and reliable way). It's the difference between me saying "My Cubbies can kick the Yankees' asses," and you saying "Wanna bet?" You better believe that I like the Cubs, but there's no way in Hell that I'm going to risk my finances on them against the Yankees.
Same goes for sites that use Stronghold or IIS/Commerce Server versus "JoeBob's Pet Cat Page." If JoeBob's website is unreliable, it's a pain in the arse, but that's about it. If Dell or Amazon.com's site goes down for even a couple of hours, that's millions of dollars lost. You can look at it as a case of "Who do I think is neat?" versus "When my financial well-being is on the line, who do I actually trust?"
Or, to phrase it in the spirit of the original example: It's all about talking the big talk versus putting your money where your mouth is.
Cheers,
ZicoKnows@hotmail.com
I'm looking at the Dec. 6, 1999 issue of PCWeek, page 58. Big ad from Microsoft, which says:
"Surprisingly (to some), Microsoft has not gone the way of the brontosaurus. The proof? Microsoft(R) Windows(R) DNA, our comprehensive platform for easily building distributed Web apps today. More proof? There are more Web sites (including Ask Jeeves, drugstore.com, Dell.com, and Nasdaq(R)) running on our platform than any other, including Sun Solaris *.
*Netcraft Survey (September 1999).
What the hell are they talking about? Anyone can see Apache ran on twice the number of host names as Microsoft in September (and today). Maybe Microsoft likes to think of "platform" as being an operating system plus web server etc. Well, not only is the operating system not provided by Netcraft, but on the page where Netcraft summarizes "platform groupings", the groupings are NT, Apache, Roxen, and Macintosh. So Netcraft certainly considers Apache a platform.
I cannot think of any reasonable explanation for Microsoft's claim. IANAL, but I think the ad is highly misleading. And at the very least, they should be made to remove the footnote to Netcraft. Thoughts?
Netcraft also surveys SSL servers on the Internet, and there the numbers are very different: Microsoft's 37 percent is almost twice that of Netscape's or Stronghold's share, and Sun Web Server doesn't even make the list.
.6272. . 30.66
.110. . .0.54 .16. . . 0.08
Stronghold? Uhhh... I have Apache. I have SSL on my Apache. I do not have Stronghold. Just because Stronghold is a hardened Apache, doesn't mean it's the only hardened Apache.
I only have this sample to go by, but right off the bat you can see a how different the plot can be if you don't cut part of the story:
Server..........Sites...Percentage
Microsoft-IIS.
Stronghold. . . 3277. . 16.02
Apache. . . . . 3062. . 14.97
Apache-SSL-US.
Apache-us-ssl.
Apache-SSL. . . 12. . . 0.06
(Aaaarrrgh! Why can't we use tables or 'pre' tags?)
Let's see... (whips out calculator) that makes:
31.67% using an Apache with SSL (6477 sites) against Microsoft's 30.66% (6272 sites). And that's only the servers which are recognizably Apache derivates (I'm sure some people here could add a few to the list)
Sure Apache only has a small lead, but that's a lot different than only half as many! Look a little more deeply next time.
Chris
San Francisco values: compassion, tolerance, respect, intelligence
Go to http://www.netcraft.com/whats/ and enter your hostname:port. (The port, of course, defaults to 80.) You'll be revisited by their bot monthly.
Gates' Law: Every 18 months, the speed of software halves.
Number of pages is a meaningless statistic: most large sites have a fair amount of dynamic content nowadays. Of course, one can argue that the number of hostnames is also a meaningless statistic, but I challenge you to find a better one.
Gates' Law: Every 18 months, the speed of software halves.
Yeah, I took the data from a bunch of Netcraft surveys and put it in a cluster. I call it a "chart". :)
Gates' Law: Every 18 months, the speed of software halves.
Like a loosing politician hearing a poll, you doubt the data before you, and yet Apache continues to be the leader.
/.'d:
I quote this data from an article previously
"Though Apache is obviously in wide use, this statistic overstates its popularity. Why? Because it measures Apache's market share based on the number of unique domain names, as opposed to the number of unique servers. Apache, popular among ISPs because of its price (free) and unlimited configurability, can host hundreds or thousands of domains on a single server.
Netcraft also surveys SSL servers on the Internet, and there the numbers are very different: Microsoft's 37 percent is almost twice that of Netscape's or Stronghold's share, and Sun Web Server doesn't even make the list. Because SSL servers are more likely to be real business servers (as opposed to those hosting "Great Doghouses of the South" or other home-page marginalia), Netcraft's survey is a telling indicator of Microsoft's presence in the market."
Given MSs very small share, it's no wonder those numbers have gone up, 'double not-much' is still not much.
Regardless of how you cook (or ignore) the data, MS's share cannot be described as "small," unless you are involved in some serious wishful-thinking.
Neopets - the best free game on the Int
The source of the second issue you brought up is mentioned in the article referenced just a little while ago here on Slashdot. The Netcraft survey is counting unique domain names, not number of servers according to the article.
The net will not be what we demand, but what we make it. Build it well.
--
WorldServe Consulting
Who's quality was all too catchy.
Our good old friend Bull
His milk did he spill
When he heard of this market share banshee.
Well, at least I tried. :)
There's no reason for a sig here.
Nope. Even the webmasters of a site have a hard time getting truly reliable stats. What happens if someone decides that they want to up the hit counter on a site?
Now just imagine trying to figure out if a webmaster is lieing. You can't.
The only way to get reliable stats on a web site is to actually go and monitor users. Say by monitoring their http requests at major routers. But not only are their privacy issues here but how are you going to do that? You would need the co-operation of the ISPs. And even then you might not be right due to HTTP requests going through other routes.
Even the netcraft HTTP requests can be faked. But I don't think many people would bother...
And with those netcraft studies people are still complaining that "the big sites" don't use Apache and instead use Microsoft. And many of the sites running on Apache are actually just small-time sites getting very few users.
You can't win...
I keep hearing this argument that even though apache has more sites, Microsoft controls the ecommerce site market. Is this true? Also, I've heard that the 5 million site count is biased because most ISPs use apache and therefore every little piss-ant website counts as a apache. Please fill me in on the straight story so I know how to respond to the FUD. Thanks all!
[frisco@hormiga frisco]$ telnet www.uu.net 80
Trying 208.243.117.123...
Connected to www.uu.net.
Escape character is '^]'.
HEAD / HTTP/1.0
HTTP/1.1 200 OK
Date: Mon, 27 Dec 1999 01:00:28 GMT
Server: Apache/1.3.9 (Unix)
Connection: close
Content-Type: text/html
Connection closed by foreign host.
[frisco@hormiga frisco]$
point being, www.uu.net is running apache.
-f
www.blackant.net
is here , where you can see that, even if the market share varies, they are all increasing in total use.