Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
2003: Year of Apache

By John Chamberlain in Internet
Sun Jan 11, 2004 at 03:58:35 AM EST
Tags: Technology (all tags)
Technology

Netcraft's numbers for the new year are in. The trend graphs tell a story: 2003 was the Year of Apache. If Time magazine had a server-of-the-year award the cover would be featuring a feather. Since October 2002 market share has grown from 53% to 64%, a 20% gain while Microsoft IIS, its nearest competitor has shrunk from 36% to 24%, a 33% decline. The change in server totals was even more dramatic. Apache HTTP Server increased from about 20 million to 32 million (+60%) while all other competitors remained flat. What are the driving factors behind this trend?


Yesterday I wrote a diary entry on Java versus .NET in which I concluded Java was leaving .NET in the dust. Predictably I got some flak from Microsoft supporters, but even more surprising were claims that Java could not stand up to Microsoft's "money and market share". So I checked the server statistics you see above. Obviously Microsoft's market share is not so invulnerable after all. IIS is sinking like a rock. But is it just because of Java? No.

Windows Distribution Hikes Apache HTTP

The numbers show Apache with total server gain triple that of market share gain, which indicates broad-based adoption. Smaller companies and individuals are switching to Apache faster than the big sites. One of the reasons for this is that Apache's support for Windows increased last year and it became a lot easier to run the HTTP server on a Windows machine. For developers like me who were already using Apache at work this was an incentive to make the switch.

The Security Factor

In September 2001 Nimda was a crushing blow to IIS users. Even so IIS market share quickly rebounded in the first half of 2002. Nevertheless the damage had been done. IIS operators who require 24/7 reliability had their confidence shaken. When Microsoft followed up with one security patch after another the distrust was compounded with an administrative burden. You can almost read the security patch effect as IIS market share dragged down over the next year and a half.

Java Trickle Down and Trickle Up

As Java grows it has a dual effect that benefits Apache. Large corporate sites that are adopting J2EE often switch to Apache. Smaller companies then imitate the larger ones and their developers start installing Apache on their home machines. Apache trickles down.

It also trickles up. Java has become the dominant language in education, replacing C and Pascal. Every year a new crop of students graduates having taken all their computing courses in Java. In many cases their schools also use Apache. These new entrants to the market are not necessarily making buying decisions for companies, but they have a distinct trickle-up effect.

The Technology Factor

The factor I emphasized in diary entry was the technology factor. Java and J2EE are by any measure more technologically expansive than C#/.NET. There are more libraries and greater support for things like mainframes and clustering. In the last year this disparity has intensified. Blame it on the Java community process or on cross-platform ability or on public source code and standards. For whatever the reason Java technology is expanding faster than .NET and that is driving strategic decisions at companies to switch to the Apache/Servlet model of operation.

Analysts Must Eat Words

This change was definitely not predicted. In 2002 market research firms were widely predicting not an Apache explosion, but a .NET explosion. These contrarians include Ovum (Microsoft's .NET's far ahead of rivals), Gartner (Gartner Predicts Return to Rich-Client Application Deployment Will Be Led by Microsoft's .NET), Timothy Dyck (.NET most impressive technology of 2002), and Meta Group (Java Keeps Brewing, But .Net Looms), among others.

____________________________________________

I admit that a year ago I was skeptical of the Java/Apache development scheme to prevail long-term against IIS and other competitors, but 2003 has demonstrated that the momentum has decisively shifted to the open-source side.

- John Chamberlain

Sponsors

Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure

Login

Related Links
o Netcraft
o trend graphs
o Apache HTTP Server
o Java versus .NET
o Microsoft' s .NET's far ahead of rivals
o Gartner Predicts Return to Rich-Client Application Deployment Will Be Led by Microsoft's .NET
o .NET most impressive technology of 2002
o Java Keeps Brewing, But .Net Looms
o John Chamberlain
o Also by John Chamberlain


Display: Sort:
2003: Year of Apache | 88 comments (81 topical, 7 editorial, 1 hidden)
+1 Tech (1.11 / 17) (#2)
by blackpaw on Sat Jan 10, 2004 at 05:09:26 PM EST

I don't agree with the java stuff, but thats not the point for voting anyway is it ?

Good to see a tech article after the endless fictions/political/porn trolling ...

A pity (2.00 / 15) (#5)
by ucblockhead on Sat Jan 10, 2004 at 06:34:19 PM EST

Because C# is a better language than Java. It's got more features, allows a coder to produce cleaner code and the .NET APIs are better organized and more functional than the Java equivalent.

Hopefully, Sun will improve Java and include some of the useful features that are in C# today. (Like properties, delegates, boxing, foreach, etc.)
-----------------------
This is k5. We're all tools - duxup

Java 1.5 (2.40 / 5) (#6)
by luthe on Sat Jan 10, 2004 at 08:44:13 PM EST

auto boxing, and a for_each are both coming in 1.5. Off the top of my head 1.5 also has an expanded set of threading related classes, generics. If you poke around sun's site I think there is a beta jdk that supports some of this stuff.

[ Parent ]
Generics will be nicer in c# (none / 2) (#18)
by The Central Committee on Sun Jan 11, 2004 at 02:07:19 AM EST

I think java generics is a wrapper around Object at the compiler level, while c# will be able to generate code at the runtime stage for specific types like primitives.

You personaly are the reason I cannot believe in a compassionate god, a creature of ineffable itelligence would surely know better than to let someone like you exist. - dorc
[ Parent ]

generics (none / 0) (#59)
by ucblockhead on Mon Jan 12, 2004 at 01:54:17 PM EST

Apparently they've built support for generics directly into the CIL.
-----------------------
This is k5. We're all tools - duxup
[ Parent ]
Mono. (none / 3) (#24)
by damiam on Sun Jan 11, 2004 at 09:54:07 AM EST



[ Parent ]
Ada 95 is already here (none / 1) (#69)
by bsavoie on Tue Jan 13, 2004 at 10:21:41 AM EST

Why use Java as a measure of what is best? Remember when Basic was taught in schools? Our education system is in decline, as is our whole culture.

Complex 'natural' software systems must be multi-threaded, Ada's protected objects makes that easy. Sure it might be hard to learn how to program but how hard is it to debug bad programs? I have spent many years debugging assembly language programs, and now that we have CPU power, Ada is a great way to go. It is also open source, it comes with SuSE and Debian. With GTK bindings and Glade it is easy to use and cross platform portable.

Bill Savoie www.dyad.org
May Peace and Love be your path
[ Parent ]
Client vs Server (2.63 / 11) (#7)
by swr on Sat Jan 10, 2004 at 09:05:11 PM EST

It seems to be taken for granted that Microsoft is the market leader who can crush all others like bugs. This is true on the desktop, but everywhere else MS is just another vendor.

Expect to see C# usage rise, but not at the expense of Java. C# will be the language of choice for Windows apps, displacing C++. C# carries the risks associated with vendor lock-in[*], and most server-side folks understand what that means, being accustomed to an actual market with actual choice.

MS could leverage their desktop position (one of the links at the bottom mentions "rich client" apps) but their control over the desktop isn't what it used to be, due largely to their own roadmap. To leverage the desktop, they need technologies on the desktop that depend on their server, and those technologies need to be ubiquitous. But MS isn't going to release their next OS until probably 2006 or so, and they've said that there aren't going to be any new versions of IE except through the new OS, which means they don't have anything to leverage and won't for quite some time.

[*] Yes I know about Mono, but you can bet most C# code is going to be targetted to Microsoft's .NET implementation and won't even be tested - let alone actually work - with anything else.



not another vendor (2.80 / 5) (#19)
by martingale on Sun Jan 11, 2004 at 04:31:29 AM EST

What makes Microsoft different from other vendors is precisely their willingness to leverage their monopoly. If Gates wanted to, he could piss away 20 billion dollars as a loss leader to capture the server market. Think about that. Except for open source, which doesn't need to make a profit, ever, all possible competitors to Microsoft would lose if Gates woke up one day and announced "I have a dream".

You can argue that investors wouldn't be too happy about pissing away 20 billion, but they couldn't do a whole lot about it, and they'd probably still lap it up as the holy truth from Gates' mouth.

You can argue that companies still prefer higher quality tech over Microsoft technology, but remember that many of the top Microsoft software titles were simply purchased and rebranded. Microsoft can buy any competitor company, or simply hire their best talent by making an offer no one can refuse, or make their software bizantinely incompatible, or any number of other things you can imagine for 20 billion in cash.

You can argue that governments would step in, but we've seen where that went the last time.

And don't even think about grass roots consumer efforts leving Windows in droves.

The fact is that unless and until their cash reserves and monopoly have been seriously eroded (90% or more), they're not "just" another competitor. And remember, I meant cash reserves, not value on paper. Think about what they could raise from investors if they needed to.

[ Parent ]

Yes, but it's irrelevant... (none / 2) (#61)
by trimethyl on Mon Jan 12, 2004 at 03:09:49 PM EST

Microsoft has truly failed to capture the enterprise market. By far, the UNIX servers being retired are being replaced with Linux systems, not W2k/W2k3 systems.

Microsoft fails to understand the enterprise computing market. Even should their software achieve true enterprise-class security and reliability, they still don't know how to market their wares to this class of customer. And now that the desktop has become a commodity item, the enterprize market is the market where most of the IT dollars will be spent.

The Microsoft monopoly isn't going to go away, but it will become irrelevant. Writing desktop operating systems isn't going to be a profitable enterprise for very much longer - especially now that PC's are being sold for less than $200. In order to stay profitable, Microsoft is going to have to fight and win an uphill marketing battle against the likes of Compaq, Sun, and IBM, who already have a well-established, loyal customer base in the enterprise market - not to mention greater cash reserves.

It's kind of ironic that Red Hat, et al, are poised to destroy Microsoft's market for operating systems in much the same way that Microsoft destroyed Netscape's market by giving away IE.



[ Parent ]
2003: Year of Windows XP (1.03 / 26) (#9)
by Linus Torvald on Sat Jan 10, 2004 at 11:08:53 PM EST

That's right, Google's figures show that Windows XP is leading the pack of OS, with a 42% of overall OS market share compared to a paltry 1% for Linux.

http://www.google.com/press/zeitgeist_nov03.html

Hot gay sex now

That's desktop share (2.80 / 5) (#13)
by John Chamberlain on Sat Jan 10, 2004 at 11:57:02 PM EST

That's desktop share. Servers are a different story.

[ Parent ]
Read the fucking link (1.00 / 6) (#25)
by CtrlBR on Sun Jan 11, 2004 at 10:29:15 AM EST

The pointed page does not have any occurence of the word "linux" nor of the word "windows". YHBT.
If no-one thinks you're a freedom fighter than you're probably not a terrorist.
-- Parent ]
yes it does (2.20 / 5) (#28)
by mikpos on Sun Jan 11, 2004 at 10:38:03 AM EST

When reading a page with images, it's important to keep in mind that images exist. Especially pie charts.

[ Parent ]
Well, (none / 2) (#40)
by abulafia on Sun Jan 11, 2004 at 05:58:59 PM EST

... if it doesn't display in Lynx, it isn't a real web page. Damn, but I wish people would realize that tables are a Netscape proprietary extension, and think about people using Mosaic... And what's this with opening multiple connections to a server? It it hammering my box! And Gopher is much more appropriate for most things people are putting up for http.

Wait, this is _2004_?

[ Parent ]

really shows how ignorant you are (none / 2) (#44)
by xutopia on Sun Jan 11, 2004 at 07:40:50 PM EST

The Zeitgeist is for hits on Google. I don't know about you but I don't use my servers to browse the web but to serve pages up to people.

If you can't comprehend something as simple as that maybe you shouldn't be posting.

[ Parent ]

A question (2.40 / 10) (#10)
by Kasreyn on Sat Jan 10, 2004 at 11:23:01 PM EST

In the third graph on the trends page, "Market Share for Top Servers Across All Domains", something called NCSA was doing great with about a 52% market share in 95, then in 96 tanked completely. What was this product? I've never heard of it.

Every year a new crop of students graduates having taken all their computing courses in Java.

Ugh. Now I'm going to have nightmares, thanks. :-(


-Kasreyn


"Extenuating circumstance to be mentioned on Judgement Day:
We never asked to be born in the first place."

R.I.P. Kurt. You will be missed.
Apache patched NCSA (3.00 / 8) (#12)
by John Chamberlain on Sat Jan 10, 2004 at 11:55:02 PM EST

NCSA's reference server was the standard before Apache appeared. Apache was in fact based on NCSA's server but had many patches/improvements on it (a patchy server, get it?)

Java may not be perfect but would you rather they were learning Pascal? I hated Pascal. Talk about black leather bondage. "You vill fully type everyzing..."

[ Parent ]

Not Pascal either. C. Note: not C++. C. -nt (none / 1) (#15)
by Kasreyn on Sun Jan 11, 2004 at 12:24:44 AM EST

nt could mean Nanite Terminators, Nice Tits, or Nissin Top, but it actually means NO TEXT.
"Extenuating circumstance to be mentioned on Judgement Day:
We never asked to be born in the first place."

R.I.P. Kurt. You will be missed.
[ Parent ]
why? (none / 2) (#26)
by speek on Sun Jan 11, 2004 at 10:30:47 AM EST

You'd prefer that new programmers coming from CS degrees continue to emphasize performance over security, performance over bugs, performance over maintainability? I hope in 10 years virtually all my user apps are written in high-level, extremely cross-platform languages like java/python/perl/ruby instead of C. The apps will be better, have more powerful features, be improved upon more rapidly, and run on a Linux machine.

--
al queda is kicking themsleves for not knowing about the levees
[ Parent ]

Why ? (none / 3) (#27)
by Simon Kinahan on Sun Jan 11, 2004 at 10:31:02 AM EST

I don't think C is a very good language to teach computer science. Although it is much simpler than C++, there is far too much detail that a C programmer has to worry about that obscures the essentials of techniques, algorithms and data structures (which are surely what computer science is meant to be about).

Java doesn't seem like a bad choice. Although the language has many irritating features, these do not particularly get in the way of teaching CS. The only thing that might be problematic is that Java programs pretty much have to be object oriented, and this might confuse beginners, and since CS projects tend to be "programming in the small", it is likely to encourage bad habits.

Simon

If you disagree, post, don't moderate
[ Parent ]

Excuse me? (none / 2) (#30)
by bigchris on Sun Jan 11, 2004 at 11:19:40 AM EST

No way! you obviously don't like pointers... I positively love them! C is a fantastic language to teach.

---
I Hate Jesus: -1: Bible thumper
kpaul: YAAT. YHL. HAND. btw, YAHWEH wins ;) [mt]
[ Parent ]
I have no problem with pointers (2.83 / 6) (#31)
by Simon Kinahan on Sun Jan 11, 2004 at 11:51:08 AM EST

I don't have any religious attachments to language features these days. I'm to old for that. The question isn't what you love or hate. Rather, it is what feature set is best for for teaching computer science to undergraduates, with a wide range of different levels of ability and knowledge.

In my experience (I've spent some time teaching inexperienced programmers about OO and basic computer science), those with no prior knowledge find learning C (and moreso C++) hard to deal with, partly because it is hard to read, and partly because the programming model requires a prior knowledge of computer architecture. Since teaching a programming language is usually only a secondary objective of first year computer science courses (and even less so of industrial training courses), it is better to pick one that does not have these problems.

Simon

If you disagree, post, don't moderate
[ Parent ]

Personally... (none / 0) (#85)
by skyknight on Sun Jan 18, 2004 at 09:13:51 AM EST

I don't have much experience with teaching, but I have a rather broad experience with using different languages. I am conversant in C, C++, Perl, Java, and Python, and comfortable in both LISP and Prolog. I believe strongly that all of these have their niche, and all serious software people should possess at least a minimal parlance in each.

As for a good teaching language, my vote goes to Python and here's why...

First, it's interpreted. This is a huge boon to students who are trying to learn new concepts for the same reason that it is a huge boon to engineers doing rapid prototyping. In experimentation, it's very important to have a fast turnaround time in the code/run/debug/repeat cycle. A compiled language makes you sit around while the compiler churns out object code, all for what, an execution speed performance boost? That is the most useless thing in the world when teaching students computer science. Their program is apt to only run for a for a few fractions of a second. Furthermore, many of the novice's attempts at execution will result in a syntax error, and having to wait around for an object code compiler is surely a waste there.

Second, it has a very clean, minimalist syntax. It has decent libraries that make it a good language to use in real projects, but they are not embedded in the core language syntax, one of the chief double edged swords of Perl. Furthermore, it is easy to write both OO and procedural programs in Python, a flexibility that Java does not offer.

I am, however, a strong believer that people should learn C as soon as possible, as well as dabble in a little bit of assembler, if just for pedagogical purposes. Understanding how function calls, argument passing, and pointer dereferencing work under the hood is a very valuable experience. I am a big proponent of high level languages for much of every day work, but I remember having to write my own OS level process context switching code as one of the more fascinating experiences of my undergraduate years.



It's not much fun at the top. I envy the common people, their hearty meals and Bruce Springsteen and voting. --SIGNOR SPAGHETTI
[ Parent ]
pointers (2.80 / 5) (#32)
by Work on Sun Jan 11, 2004 at 12:10:39 PM EST

for the first year CS student pointers are difficult to grasp, even for the students who catch on to the rest quickly.

You really need to understand the memory architecture (usually by taking a course which teaches the fundamentals of the hardware, as well as some assembler) before diving full on into pointers.

Java is quickly taking over from C and C++ at major universities as the language to teach. Alot of it has to do with students often spend more time debugging and decyphering meaningless compiler errors than actually learning their data structures and overall concepts. Java occasionally throws up stupid errors unrelated to the real problem, but its an order of magnitude improved over C++.

[ Parent ]

should use Python powerful (none / 0) (#35)
by modmans2ndcoming on Sun Jan 11, 2004 at 01:38:29 PM EST

as a way to learn the programming process, Python would make a very and easy to learn language.

[ Parent ]
I disagree. (none / 0) (#54)
by tkatchev on Mon Jan 12, 2004 at 06:11:14 AM EST

I think the most important thing for beginner-level programmers to absorb is discipline, lots and lots of it.

I vote for Scheme, or maybe some academic language that is even more anal.


   -- Signed, Lev Andropoff, cosmonaut.
[ Parent ]

Python promotes good programming techniques (none / 0) (#58)
by modmans2ndcoming on Mon Jan 12, 2004 at 10:56:25 AM EST

because white space matters in python, and the way the White space was constructed makes the programmer create a program with very good readability.....there is nothing loose about Python.

just because it has dynamic memory allocation and dynamic binding of variables does not mean you do not learn discipline.

[ Parent ]

Yes, but. (none / 0) (#66)
by tkatchev on Tue Jan 13, 2004 at 06:36:58 AM EST

You'd be surprised how contrived and convoluted Python programs can be.

   -- Signed, Lev Andropoff, cosmonaut.
[ Parent ]

Formatting (none / 0) (#80)
by mohaine on Thu Jan 15, 2004 at 12:41:46 PM EST

because white space matters in python, and the way the White space was constructed makes the programmer create a program with very good readability.....there is nothing loose about Python. Formatting is best left to a program, not the programmer. IMHO formatting is probably the least important thing a new programmer needs to worry about. It should be automatic. While I like Python, I find the whitespace reqirements/formatting to be a little weird. For instance, why do you need a : before a if block if whitespace is the desciding factor? You can tell where the block of code starts from the whitespace. A begin/end or {/} would be easier then a :/whitespace and be machaine formattable.

[ Parent ]
Formatting (With readable formatting:) (none / 0) (#81)
by mohaine on Thu Jan 15, 2004 at 12:43:03 PM EST

because white space matters in python, and the way the White space was constructed makes the programmer create a program with very good readability.....there is nothing loose about Python.

Formatting is best left to a program, not the programmer. IMHO formatting is probably the least important thing a new programmer needs to worry about. It should be automatic.

While I like Python, I find the whitespace reqirements/formatting to be a little weird. For instance, why do you need a : before a if block if whitespace is the desciding factor? You can tell where the block of code starts from the whitespace. A begin/end or {/} would be easier then a :/whitespace and be machaine formattable.

[ Parent ]
You laugh... (none / 0) (#60)
by unDees on Mon Jan 12, 2004 at 02:45:42 PM EST

...and rightly so, but Scheme is indeed inflicted on many first-year college students.

Your account balance is $0.02; to continue receiving our quality opinions, please remit payment as soon as possible.
[ Parent ]
I'm not laughing. (-) (none / 0) (#67)
by tkatchev on Tue Jan 13, 2004 at 06:37:14 AM EST


   -- Signed, Lev Andropoff, cosmonaut.
[ Parent ]

Not really. (none / 0) (#84)
by bigchris on Sun Jan 18, 2004 at 08:21:27 AM EST

How hard is it to declare a pointer to a block of memory, then use malloc and free to access it?

You hardly need much of an understanding of memory architecture at all, let alone the fundamentals of hardware!

Meaningless compiler errors are hardly a problem with C's pointers. It's a problem with the compiler.

---
I Hate Jesus: -1: Bible thumper
kpaul: YAAT. YHL. HAND. btw, YAHWEH wins ;) [mt]
[ Parent ]

You are positively the CS macho dude. (2.42 / 7) (#34)
by tkatchev on Sun Jan 11, 2004 at 01:04:58 PM EST

I bow down before your massive, intellectually immense mastery of pointers and tivial high-school level programming exercises. Word up.

   -- Signed, Lev Andropoff, cosmonaut.
[ Parent ]

Uh, and that was necessary because? (none / 1) (#83)
by bigchris on Sun Jan 18, 2004 at 08:18:55 AM EST

I bow down before your massive, intellectually immense mastery of sarcastic wit. Word up.

---
I Hate Jesus: -1: Bible thumper
kpaul: YAAT. YHL. HAND. btw, YAHWEH wins ;) [mt]
[ Parent ]
He's obviously a VB programmer (n/t) (none / 1) (#86)
by Skywise on Sun Jan 18, 2004 at 12:06:39 PM EST



[ Parent ]
Is Egg Troll still around? (none / 0) (#87)
by bigchris on Tue Jan 20, 2004 at 02:32:49 AM EST

If so, then I wish to talk about O/Ses with him. In VB. tkatchev would make a valuable contributor to our new O/S: Vinux.

---
I Hate Jesus: -1: Bible thumper
kpaul: YAAT. YHL. HAND. btw, YAHWEH wins ;) [mt]
[ Parent ]
What's wrong with BASIC? (none / 0) (#56)
by onealone on Mon Jan 12, 2004 at 07:44:27 AM EST

I think all teaching of CS should be done with BASIC, and I don't mean VB either. Good old fashioned Sinclair 48k BASIC.

[ Parent ]
mod_perl (2.93 / 16) (#20)
by brainrain on Sun Jan 11, 2004 at 07:06:59 AM EST

Under 'The Technology Factor' I would add a little bit about mod_perl and PHP, their growth in popularity and general acceptance as a defacto in web application development. Developers are lured to Apache due to its tight module-integration scheme, which has long been a mainstay of web application development.

--
Kleptotherapy - Helping those who help themselves
This analysis should look at active sites (2.93 / 16) (#21)
by cabalamat on Sun Jan 11, 2004 at 07:56:05 AM EST

This analysis includes all sites, when I think it would give a more accurate picture if an analysis of active sites only were done.

I've done that analysis, and the results were that during 2003, considering new sites, Apache was chosen four times more often than IIS. The actual figures were for Apache 3.66 million new sites (77.54%) and for IIS only 0.9 million new sites (19.06%).

how many website owners actually choose? (none / 2) (#38)
by gps on Sun Jan 11, 2004 at 05:01:12 PM EST

one nit pick, don't most website operators just use whatever hosting their ISP provides (until you get to the huge sites where they use their own machines)?

in these cases (ie: most sites i suspect) it is the ISP that made the choice.

[ Parent ]

You said it. (none / 2) (#41)
by Dr O on Sun Jan 11, 2004 at 06:44:43 PM EST

> in these cases (ie: most sites i suspect) it is the ISP that made the choice. <

Yes, exactly. And they use Apache as it simply isn't the royal pain in the ass deploring, configuring and maintaining which is ISS. Apache isn't configurable in a pointy-clicky way. If you have to set up and maintain LOTS of web servers, Apache definitely is the way to go.

edit httpd.conf
kill -hup 12345

done.

Need to add a setting or feature to all of your servers? Come up with a very small shell script(TM), pipe it via ssh to your webservers and update them in a matter of minutes instead of days. This is why MS loses in spite of the not all that bad idea which is .NET.
I love deadlines. I especially like the whooshing sound they make as they go flying by. - Douglas Adams
[ Parent ]

The reason is probably also parked domains (2.50 / 4) (#48)
by klugegod on Mon Jan 12, 2004 at 12:53:44 AM EST

Whether active or non-active sites are considered, both figures include large numbers of domains parked by people who just wanted to register a domain name and resell it in the future at a higher price- one of those sites that just have a "buy this domain" link.

Recently, netcraft also reported that a number of registrars that provide such services switched to Apache this year.

Thus many parked domains, which represented a good share of newly registered and old domains, are hosted on the same server, moreover by just a handful of companies.

[ Parent ]
Methodology (none / 1) (#74)
by swr on Wed Jan 14, 2004 at 03:15:48 PM EST

No. The "active sites" stats would not count a bunch of parked domains as more than one site. That, and giving less weight to sites that are on the same IP as >100 other sites, is the difference between the regular count and the "active sites" count.

See here: How many Active Sites are there?

[ Parent ]

An unanswered question (1.66 / 12) (#22)
by The Turd Report on Sun Jan 11, 2004 at 08:24:23 AM EST

Would you have to be Kreskin to know this?

Linux (1.18 / 16) (#23)
by ylikone on Sun Jan 11, 2004 at 09:51:15 AM EST

Apache rules because Linux is taking over. Microsoft's days are numbered!

Agreed! (none / 3) (#46)
by jesusreligion on Sun Jan 11, 2004 at 10:26:25 PM EST

Ditto!
LC
[ Parent ]
me too (none / 0) (#76)
by Rahaan on Wed Jan 14, 2004 at 11:27:26 PM EST

I agree


you know, jake.. i've noticed that, since the tacos started coming, the mail doesn't so much come as often, or even at all
[ Parent ]
Marginalization of IIS (1.83 / 12) (#29)
by kondor on Sun Jan 11, 2004 at 11:06:50 AM EST

Extrapolating this trend means that within a few years, IIS will be dead.

Re: Marginalization of IIS (2.80 / 5) (#51)
by peanutbadr on Mon Jan 12, 2004 at 03:00:01 AM EST

And 10 years from now, IIS will have NEGATIVE market share.
-peanutbadr--
[ Parent ]
Verily so. (1.60 / 5) (#55)
by tkatchev on Mon Jan 12, 2004 at 06:12:07 AM EST

And in a few more years after that, IIS will have negative marketshare.

Yay for science.


   -- Signed, Lev Andropoff, cosmonaut.
[ Parent ]

Absolutely (none / 1) (#72)
by veldmon on Wed Jan 14, 2004 at 10:14:08 AM EST

And then a few years down the road, IIS will enter into negative market share territory.

You heard it here first.

[ Parent ]

oh man! (none / 1) (#75)
by Rahaan on Wed Jan 14, 2004 at 11:27:10 PM EST

and in a few years, IIS will have negative marketshare!!


you know, jake.. i've noticed that, since the tacos started coming, the mail doesn't so much come as often, or even at all
[ Parent ]
Excuse me, sir, (none / 0) (#77)
by Rahaan on Wed Jan 14, 2004 at 11:28:23 PM EST

Can you tell me why you haven't posted any serious commentary in over a month?


you know, jake.. i've noticed that, since the tacos started coming, the mail doesn't so much come as often, or even at all
[ Parent ]
Because of threads like this? (none / 0) (#78)
by Rahaan on Wed Jan 14, 2004 at 11:28:48 PM EST

yes.


you know, jake.. i've noticed that, since the tacos started coming, the mail doesn't so much come as often, or even at all
[ Parent ]
and yet... (2.50 / 10) (#33)
by Work on Sun Jan 11, 2004 at 12:29:37 PM EST

netcraft tells me my server is running apache on bsd...

I'm running resin on winxp.

you sure? (none / 1) (#43)
by jacoplane on Sun Jan 11, 2004 at 07:01:36 PM EST

it seems to identify it correctly...

[ Parent ]
interesting (none / 1) (#45)
by Work on Sun Jan 11, 2004 at 07:49:17 PM EST

They seem to have fixed it. Those other 3 are completely incorrect. American Ski Company? WTF?

[ Parent ]
The importance of sites (2.33 / 6) (#36)
by svampa on Sun Jan 11, 2004 at 03:03:50 PM EST

I read a post about Apache vs IIS

"Apache, the leading server for sites doing nothing"

Is it true? are most Apache sites hosting personal pages that almost none hits? It's not the same Slashdot or hotmail than my personal site with 200 hits/month and 50 e-mails/day.

Is there any way to know which are the more active servers? that is, those which have more trafic, or deliver more messages, or something like that?



yeah (none / 1) (#39)
by Work on Sun Jan 11, 2004 at 05:53:14 PM EST

alot of linux distros have apache running by default, even if the owner isn't using it.

I always take netcraft with a big grain of salt. Like I mentioned in an earlier post, they often misreport things depending on how your network is set up.

[ Parent ]

What? (none / 0) (#65)
by Vesperto on Mon Jan 12, 2004 at 06:49:42 PM EST

What kind of SO would run a webserver by default? If you chose it in the instalation process it'll be installed and, in most distros, you'll be warned that it'll be active (or, in decent ones, asked if you want that).

As for sites (parent post), hiting 404s of big sites is a way :)

If you disagree post, don't moderate.
[ Parent ]

Using squid cache logs (none / 0) (#53)
by cpghost on Mon Jan 12, 2004 at 05:35:27 AM EST

Perhaps a big squid cache at big ISPs could reveal the relative amount of hits/traffic/... of IIS and Apache servers?
cpghost at Cordula's Web
[ Parent ]
Doesn't contradict predictions (2.62 / 8) (#37)
by BushidoCoder on Sun Jan 11, 2004 at 04:02:19 PM EST

I'm not surprised at all by these results, and I also don't think they that adequately demonstrate a lack of market share of IIS. I think they that demonstrate a decreasing market share of IIS amoung public internet sites.

Microsoft's .NET technologies are actually very impressive, and to my experience, are being adopted quickly as the defacto way to create private intranet sites. If you look at ASP.NET, this makes sense - While featureful and easy to develop with, its also more computationally heavy than php and perl-based sites, and the frequent use of postbacks and the viewstate would result in a significantly larger bandwidth requirement on public internet sites. Skipping that, very few organizations need public sites with ASP.NET features. When you mix that with Microsoft's track record for security this year and the relative "newness" of Win2k3 and IIS 6, this report doesn't surprise me at all.

Until mod_mono is finished and gains some performance enhancements, ASP.NET on IIS is still the best way to quickly develop rich, secure intranet clients - But if you don't need ASP.NET, IIS has nothing to offer over Apache.

\bc

Wrong Interperetation (2.85 / 21) (#42)
by MyrddinE on Sun Jan 11, 2004 at 06:58:03 PM EST

I believe you are misinterpreting the graphs. The top two graphs are indicating how many domains run on each type of server. Apache is certainly doing very well here... around 32 million domains run on Apache servers, with increases like you described.

Near the bottom however, we have a different story told. Down there is the number of SERVERS installed. At the top is, of course, Apache with about 15 million active servers. However, when you look at the marketshare changes for active servers, it's completely flat.

What is happening is that Apache servers are running MORE WEBSITES per server. In other words, the big website providers, that host 5-500 websites on a single box, are running Apache. Those big companies are getting a lot more subscribers, in leaps and bounds, as hosting prices drop dramatically. Everyone and their mother are getting a website, but not on their own server... on a big hosting farm, which runs Apache.

Of course, more servers are coming out... but the market share, server per server, is barely changing, as the last graph shows. In the last year, IIS installations rose from about 4 million to about 5 million. In that same period, Apache went from about 11 million to 15 million.

BOTH servers rose about 25% over the past year, so the market share barely changed. But Apache servers are hosting many more websites per server... Apache tends to be run by ISPs and hosting companies, because it is much better at running for colocated users, who can administer their site with a shell account and via the great web management packages written for Apache. IIS tends to be used by companies running their own server, which means fewer domains per server.

So while Apache's domain count rises meteorically, the installed server base is rising on par with IIS.

I'd like to stress that I have no particular love (or dislike) for Apache or IIS. But I don't like seeing people manipulate statistics. The graph titles are misleading, so I do not think it was intentional. I'd just like to set the record straight.

New Math? (none / 1) (#71)
by onecelt on Tue Jan 13, 2004 at 03:08:36 PM EST

Your math is a bit off. If the number of total servers goes up and Apache an IIS each have the same percent increase then Apache winds up with a bigger slice of the pie. The rich get richer and the poor get poorer. IIS went from 4:15 million to 5:20 million or 26.7% to 25% (ignoring the other servers) a 1.67 percent decrease. Where Apache went from 11:15 million to 15:20 million or 73.3% to 75%. But your math is off further I don't see how you have both IIS and Apache increases of 25%. IIS increased 1:4 or 25% but Apache rose 4:15 or 26.7%. Apache is clearly running away with the title, and for good reason.

[ Parent ]
yes, read this title (none / 0) (#73)
by chanio on Wed Jan 14, 2004 at 01:04:23 PM EST

Not only bad interpretation of statistics. Why '2003 Apache year', and not '2004'? What is wrong in OS apart from improving paid software in the long run?

There is another myth stating that the best of M$ is their greed for money. They in fact do good software and the best of it is that they make them simple.

It is like comparing an European movie with an American one. To do such simple American movies, directors need a lot of skills.

So, the only difference between intellectual good European movies and American ones cannot be money. Although it is known that simple and good always sells more.

The same should apply to OS. Apache is good programming and is obtaining what they deserve. And paid systems should improve a lot in order to get paid for something that OS might be doing better...
________________
Farenheit Binman:
This worlds culture is throwing away-burning thousands of useful concepts because they don't fit in their commercial frame.
My chance of becoming intelligent!
[ Parent ]

Apache developers need to say cool. (2.75 / 8) (#47)
by tuj on Mon Jan 12, 2004 at 12:42:27 AM EST

Recent discussion suggests that Apache developers don't want to invest the time to review and probably recode the process-per-connection handling model to allow true performance scaling.

Also, they've more or less abandoned 1.3 but 2.0 still doesn't have essential modules ready for it, most notably a mod_throttle or mod_bandwidth equivalent.

Other web servers haven't made these mistakes.  No one should rest on their heels in this kind of essential-tech-product environment.  The climate will soon be such that your choice integrated services platform (read application and web server) will be more important than your operating system.

Java open source? (2.00 / 6) (#49)
by STFUYHBT on Mon Jan 12, 2004 at 01:06:52 AM EST

I was skeptical of the Java/Apache development scheme to prevail long-term against IIS and other competitors, but 2003 has demonstrated that the momentum has decisively shifted to the open-source side.

Did I miss something? When did Java become open source?

-
"Of all the myriad forms of life here, the 'troll-diagnostic' is surely the lowest, yes?" -medham

Libraries (none / 1) (#50)
by smileyy on Mon Jan 12, 2004 at 02:09:20 AM EST

Probably the huge number of mature open-source java libraries, rather than the JDK itself.
--
...alone in suicide, which is deeper than death...
[ Parent ]
Apparently you did miss something (none / 0) (#57)
by verbatim verbose on Mon Jan 12, 2004 at 10:09:38 AM EST

look here...
Download (SCSL source)... there it is.

[ Parent ]
Open Source forever... (none / 1) (#63)
by RadiantMatrix on Mon Jan 12, 2004 at 05:27:35 PM EST

Java has been open sourced for some time -- remember that open source doesn't automatically mean FSF-approved licensing. The source is available to anyone (at no charge, even), and therefore "open". Your right to redistribute code with modifications is, however, rather restricted.

As a result, Java is "open source", but it isn't "free software" (by the FSF definition, anyhow). The two seem to get confused a lot (since free software is [nearly?] always open source).


----------
I don't like spam - Parent ]

It's pretty ironic (1.16 / 6) (#52)
by ChefQuix on Mon Jan 12, 2004 at 03:33:27 AM EST

Not that I don't love apache, however if you try and hit the apache 2.0 documentation pages between 9-5 Monday to Friday, well, let's just say it's not a very good advertisement of the quality of their software. When I have to read up on apache it's often much quicker to find a documentation mirror because it seems their software / server / pipe can't keep up with traffic. I know I could just install the documentation on my server but hey, I'm lazy. ;)


Polluting the Internet since 2003...
perceptionalism.com
Quality? huh? (none / 1) (#62)
by RadiantMatrix on Mon Jan 12, 2004 at 05:24:39 PM EST

if you try and hit the apache 2.0 documentation pages between 9-5 Monday to Friday, well, let's just say it's not a very good advertisement of the quality of their software.
I don't know how the slow response time of the doc site Mon 9-5 really makes any criticism of the quality of 2.0. Understanding that the overall network volume is high, combined with the kind of load that (likely) tens of thousands of page views per hour puts on a server, I'd say I'm impressed that the server responds at all.

If anything, the fact that you can still use the primary site during peak speaks volumes about the high quality of the Apache httpd.


----------
I don't like spam - Parent ]

See that's just it... (none / 0) (#64)
by ChefQuix on Mon Jan 12, 2004 at 05:54:50 PM EST

The site basically doesn't respond to documentation requests during the peak hours. The rest of the site seems to work fine except for the damn documentation. All I'm saying is it looks bad, especially for someone who's just thinking about heading over to apache from IIS.


Polluting the Internet since 2003...
perceptionalism.com
[ Parent ]
Just a note (none / 3) (#68)
by muyuubyou on Tue Jan 13, 2004 at 08:26:41 AM EST

Java has become the dominant language in education, replacing C and Pascal. Every year a new crop of students graduates having taken all their computing courses in Java. In many cases their schools also use Apache. These new entrants to the market are not necessarily making buying decisions for companies, but they have a distinct trickle-up effect.

I sincerely hope new graduates don't learn just one programming language... that would make them even more incompetent and useless than the latest crops of programmer wannabes coming from unrelated fields.

/shudders

Better off learning 0 (none / 2) (#70)
by drix on Tue Jan 13, 2004 at 02:17:50 PM EST

To quote one of my CS profs,
Learning a language is something you do over a weeke nd. If that's not the case, you're in the wrong field.
Any half-decent CS program teaches their students the fundamental concepts behind programming--abstraction, software engineering, OOP, data structures, blah blah blah--and leaves learning the language up to the students. At least that's how it is here at Berkeley. Anyone with a degree in computer science who tells you with a straight face, "I'm not trained in that language," should've majored in English.

[ Parent ]
One language (none / 0) (#79)
by thejeff on Thu Jan 15, 2004 at 09:52:24 AM EST

Why? It's a correct english sentence and often true.  If it's "I can't do that project, because I'm not trained in that language", you're right. But what should you say if you're asked if you have experience with a language?

The trouble is, employers are looking for language skills. Try saying in an interview, "No, I've never used that language, but I'll pick it up next weekend." Fresh out of school, that might get you somewhere, but even then it's a black mark against you.

Most lower level classes, at least, are taught in one language, for the instructor's sanity if nothing else. I'd advocate varying that across the curriculum, not teaching the language, but using different languages to teach concepts.

[ Parent ]

Agree with you (none / 0) (#82)
by james2141 on Thu Jan 15, 2004 at 08:04:00 PM EST

You're right - when you know some you can feel free to move not to be like a newbee.

[ Parent ]
it sucks (none / 1) (#88)
by rebelcan on Fri Jan 23, 2004 at 03:15:53 AM EST

i'm currently finishing up my last year in high school, and the only things i've learned in class from the teacher about computers would fill less than a byte. i feel sorry for anybody in my infotech class that thinks that what we've learned so far ( HTML,CSS,HTML,JavaScript,HTML, wee bit of perl,HTML, and we're supposed to eventually touch on Java ) will help them at all when they hit university/college, and stumble onto real languages ( we haven't done Java yet, so i'm not trashing it,and i really like perl. i'm trashing javascript, because so far, in a full-year school, we're only just started on mouseovers. ARGH.)

i don't like the way the school system is set-up because i WANT to learn the fundementals of programming in any language, but the school system seems to think it is a better idea to teach us stuff that we could pick up in a weekend from the web or a book.


=============================
God is dead -- Nietzsche
Nietzsche is dead -- God
but Zombie Nietzsche lives! -- Zombie Nietzsche
[ Parent ]

2003: Year of Apache | 88 comments (81 topical, 7 editorial, 1 hidden)
Display: Sort:

kuro5hin.org

[XML]
All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!