Main Menu
Network
Sponsor
Top 10 Sites
Partners
|
|
Top Submit newsSubscribe 
Communication | Computer Crime | Digital Audio, Video, Photo | General News | Hardware | Internet | Mobile | PDA | Security | Software | Vulnerability |
Previous articleBack to news listNext article
| Sponsored links | Want to become one of our authors and see your work published on ALLSeek.iNFO ?
| | Has Apache peaked? |
|---|
Categorie: Software
Posted: 2002-10-16 by Gmtech
Views: 576
Source: Click here
| Current Rating: Not rated
|
| | Details |
|---|
One of the great success stories of the Internet and of open source software is Apache, the dominant web server software on the Internet.
The 1.x generation of Apache has been out for many years, so it should have been a big deal when version 2.0 shipped.
But early adoption of Apache version 2.0 was minimal according to Netcraft, the authorities on what people are running on the Internet. Over 11 million active sites are running Apache, but about six months after its release, fewer than 100,000 of them are running 2.0. Why?
As popular as Apache 1.3 was and is, the design is a little long in the tooth. It's just not optimized for high-traffic, high-performance environments. The main job of 2.0 was to address these problems, especially on the Windows platform (where nobody runs Apache 1.x except to prove that it can be done). The new version did not mess with Apache's beloved, user-friendly administration interface (that's a joke, guys).
At least on paper, Apache 2.0 has a lot going for it. More extensive modularity of design at lower levels of the server has made it more flexible at adapting to various architectures. The best example of the advantages of this is that the Windows version of Apache 2.0 is much faster, at least in part, because it uses native networking facilities instead of having to rely on a POSIX (Portable Operating System Interface for Unix) layer. (One of the secrets of successful application writing on MS operating systems is that they usually give you many levels to program at, but the best applications write to the most native levels.) This modularity is implemented in the Multi-Processing Module (MPM) loaded in Apache 2.0.
Unfortunately, the changes in 2.0 necessary to implement the performance improvements were significant, and they break all of Apache's old module code. It all needs to be rewritten and--amazingly--six months after the release of 2.0, much of the job remains undone. Reports indicate that this is mainly because the release of Apache 2.0 was a misleading event: The APIs for writing Apache modules have changed frequently with incremental releases of the new Apache (there have been 30 code releases of 2.0 and the current version is 2.0.40).
In the commercial software business if a dominant product had handled a major release in this manner it would have been a major scandal. But there has been barely a peep about Apache 2.0. In fact, Netcraft's Mark Prettejohn says that the only commercial Apache company running Apache 2.x on its own site is Covalent. IBM and all the big Linux distribution companies are running 1.x. At least the Apache Software Foundation itself runs 2.0 on their site. There are some Apache 2.0 bastions out there, including Teletubbies and the Knights of the Ku Klux Klan. So Apache 2.0 adoption is broad, if not deep.
It's tempting to think that Apache 2.0 missed the boat. The vast majority of Web sites running it are hosted brochureware. The improvements in 2.0 are meaningless to them. Sites with higher-end needs are more likely to be running a higher-performance Apache alternative like IIS or Zeus already. And if an Apache site actually needs the performance improvements they would do well to treat the current version like a beta. This is how we all should think of it for now.
When we finally get a stable version of Apache 2.0 it's not clear how well it will perform relative to the competition. I have no doubt it will far outperform Apache 1.3 on intensive performance tests, but as I discussed earlier, doubts have been raised about the new process model in Apache. Many of these doubts come from competitors and other interested parties, but I think there are some valid points here and it will take some time before we know how successful the new version is.
Perhaps even a stable version won't get the ball rolling, but roll it will, eventually. As Prettejohn suggested to me, when Red Hat makes Apache 2.0 the default server installation on their Linux distribution, people will start to use it. At first I thought this incongruous; Linux administrators don't like to think of themselves as the type to blindly run whatever comes on the install disk. And I doubt Red Hat will make 2.0 the default install until these development issues are ironed out, so don't look for this any time soon.
As embarrassing as it is, that's going to be the answer for Apache 2.0. People won't be adopting it consciously; there's no need to. They'll adopt it slowly and coincidentally, as they roll out new systems. This doesn't speak well of the new version.
| | Links |
|---|
The 1.x generation of Apache
| | Syndication |
|---|
Permalink Email this
The URI to TrackBack this entry is:
http://allseek.info/news/trackback.php?id=75
| | User comments (post your comments ) |
|---|
Only registerd members can post comments and articles
|
| Previous articleBack to news listNext article
|
|
|
|
InterJOB.su
|
