Slashdot Mirror
Will Open Source Lose the Battle for the Web?
snotty writes "A well written article by Ganesh Prasad over at linuxtoday arguing that the shift towards web services has reduced the attractiveness of the current generation of Open Source web products. He talks about the market share decrease in Apache. Also mentions how .NET, Microsoft, Sun, Java, and Open Source Software fit into the picture." I think that the decrease in Apache's share is a red herring, but the bigger picture of web services is a troubling one.
Did you even bother to check with Netcraft as to what the actual cause of Apache's drop was? Are you there? Wakey, wakey?
Since you can't be bothered to follow the link, here's the quote:
What will happen next month? Apache won't be losing ground, unless Namezero had a couple of thousand domains lying around somewhere on Apache and decide to change them to IIS as well.
Namezero have thousands of domains. As for physical servers, no, they don't have many. The drop in numbers is due to a couple of free web sites that Namezero hosted being shifted.
I don't think it bodes badly for Apache. The author of the article linked was a troll, and comes from an disreputable trollish bundle of bollocks organisation. Do you think I believe their trash? As much as I eat my faeces.
imho, it may easier to create a m$ web services solution, but i don't think you'll have the same flexibility and performance
Allow me to humbly disagree. There are things that certain technologies can do and certain things they cannot. In this world, no one technology is ever 'best' or 'worst', they are either well-suited for the job or they are not.
In my example, I work for a company which was always a Microsoft shop (Microsoft partner, and sells tons of MS stuff). Guess what - our particular eBusiness solution, which was developed using the latest/greatest MS solution wasn't up to snuff. We couldn't get the flexibility out of the application we needed (even a redesign wouldn't work - VB had too many limitations), IIS either crashes on a regular basis or needs constant hotfixes to keep it secure, and SQL Server shuts itself off whenever it feels like it. And we can blame the 'expert' consultant who set the thing up initially, took off, and left me to try to support the damned mess that was left.
Enter Java. Using Tomcat as the Servlet engine, and the Velocity template engine, we have nearly completed rewriting our entire eBusiness web application, and this new model demonstrates the ability to customize both the look and feel (a custom framework using Velocity which can load customized templates per customer), and a Servlet framework which can automagically load custom code on a per-customer basis. This code is simply written to extend previous code. Both are loaded on the fly (not necessarily compiled into the main application). Speed/ease of development? We are adding new customizations and features in a day or two when they used to take a week or two to complete when using VB/ASP and COM. In other words, by doing the complete rewrite we have already saved significant amount of developer time, after that initial investment.
We are using this new app in production now, and there have been no problems, other than SQL Server still shutting itself down. This will change within the next two months, as we are moving our database to something more robust which we already use for our backend.
Still being a MS-shop, I was shocked to find out that the president of our company decided to let me go ahead and use what I want on our servers. He was getting tired of the constant security worries and instability of the thing. And yes, we are going to switch to Linux/Apache as soon as we eliminate our final few dependencies on MS-specific code.
You can accomplish anything you set your mind to. The impossible just takes a little longer.
No flame intended. But.
Microsoft's new "technologies" (which I put in quotes because what they call a "technology" is usually just a new implementation of features) always look cool in demos and books. You don't find out about the hidden "features" until you actually develop a reasonably large project.
I'll give one tiny example since it's on my mind right now. Microsoft's Component Object Model (COM) provides a function CoCreateInstanceEx, which takes a parameter indicating the DNS name or IP address of the remote host where you want a new COM object to be instantiated. Neato! It's like RPC but with type safety! As long as the remote host is configured to let your host do so, you can instantiate and use a DCOM object anywhere on the Internet or your private network. Not really a new idea, strictly speaking, but nice nonetheless.
But there's an annoying catch. You can't specify your own hostname or IP address. If you want to instantiate a COM object on your own host, you either have to pass a NULL to CoCreateInstanceEx or just call CoCreateInstance (without the "Ex") instead.
It's no big deal, but it breaks perfectly reasonable code for no reason. Gratuitous non-orthogonality.
I love Linux and other free software, and I'm a vocal Microsoft critic, but that doesn't mean I write this stuff out of ignorance. I've been programming in Windows since the 3.0 release, and have recently been spending probably about two-thirds of my work time programming various Win32 projects, from end-user database applications with CA-Visual Objects (anybody even heard of that?) to the C++ middle tier of a couple of 3-tier client/server systems. I don't hate the stuff, I just have plenty of practical knowledge of how it sucks.
COM (and DCOM--"Distributed COM") is littered with barely-documented or undocumented details like these. Like the way strongly-typed parameters can be passed by reference across process boundaries--just not from Visual Basic clients, even though COM is claimed to be language-neutral.
Microsoft development tools, I've found, generally stop doing what you expect the moment "gee-whiz!" settles into "can you make it do this?" When you read Microsoft docs or third-party books on a new "technology" it pays to be skeptical whenever the CS student inside your head says, "That's amazing! How can they do that?"
I guess I'm saying two things:
If you write a few pretty decent-sized .NET applications and they work out as well as expected, super. But until then, I'd withhold judgement.
MS bashing aside, .NET is a really great idea from a developer's standpoint, and its implementation is great.
/. vibe, but come on... This is New for Nerds, not News for Platform-Exclusive People. PHP, Perl, Java, they all have their uses--why can't people accept that, maybe, MS actually makes good products once and a while? :-)
I'm a web developer, so it's my job to stay up on the newest technologies. I recently picked up "Professional ASP.NET" by Wrox, and, about 650 pages in, I have come to the conclusion that ASP.NET is one of the most fuctional, flexible web platform I have ever seen.
Sure, PHP is just dandy... But ASP.NET adds on so much functionality, it's scary. Sure, that may be one of the reasons to shun it from a certain viewpoint--letting a MS platform have so much control--but as a developer, ASP.NET is a dream come true.
I know this goes against the general
What's a sig?
IIS does not run in a privileged (a.k.a. "kernel-mode") process. It cannot directly read/write addresses in kernel memory. As an IIS developer at Microsoft for the SMTPSVC who actively debugs IIS every day, I know very well that INETINFO.EXE is a user-mode process.
The source of your confusion may be that IIS threads have SYSTEM tokens, granted from the NT SCM which launches INETINFO.EXE. This, however, does not offer any performance boost, nor does it allow access to kernel memory space. As with any user-mode process, any attempt by an IIS thread to read/write kernel memory results in an access violation, regardless of the thread's access token.
The good performance of IIS's W3SVC is due to its asynchronous architecture that allows its threads to avoid blocking on I/O operations. Thus many I/O operations can be pending concurrently without a bloat in the number of threads, avoiding excessive context switching and VM swapping for extra stack space.
D
It looks more like 10% to me (eyeballing the first graph). What bothers me is that Microsoft went up about the same amount.
.Net bullshit, it is a bad sign.
By itself, it doesn't mean anything more than what it claims to be. Taking it along with other information such as this
Open Source will never go away. When(if) the Internet becomes controlled and commercialized, we will ignore and avoid it or go someplace else. It won't be the end of the world.
We should not really be paying attention to the Apache stats right now. We should be looking much more carefully at the Linux and BSD stats. Apache may lose some marketshare as Solaris, Irix, etc. continue to lose marketshare, but that is not a good measure of the health of open source software, or so I think. I personally would not measure Open Source health by looking into the market share of proprietary OSs.
LedgerSMB: Open source Accounting/ERP
Oh, yes they [sic] is ... JBoss. It's not part of the Jakarta project, but it integrates nicely with Apache/Tomcat. It mentioned (although only briefly) in the article, had you bothered to read it.
(Before I continue - no, I am not a Novell employee, and no, Novell is not dead) .NET is to build something around some other (non-MS) technology.
I think Ganesh is on to something here - the only way to compete with
Novell did exactly that - created a portal environment by combining Open Source, XML and Java components, while using the existing directory structure for authentication. It is still in early stages, but it does work, and it works incredibly well.
Before I see eyes rolling - yes, portals are basically warmed over web pages, but it is a corporate buzzword du jour. Your average CIO/CTO loves it, and will allow unholy amounts of money to be spent on it.
How does it work? You have an Apache/Tomcat combo sitting on one of four supported platforms (NT, NetWare, Linux or Solaris) and JDK with JNDI. (Java Naming and Directory Interface) When you log in, JNDI taps into your existing LDAP3-supported directory for authentication and profiles (Novell of course hopes you will be using NDS/eDirectory), and presents you with cutomized, pre-formated page (done with XML/XSLT) that contains your mailbox, calendar, MetaFrame apps, news, queries your databases via JDBC, etc.
And now the kicker - because Novell is using Open Source and Java components, this product is about three times cheaper than the first comparable solution - something even CFOs might appreciate.
Why I like this so much? I gives Linux foot in the door. Now I can actually *justify* using a Linux box for a web/application server instead of IIS, and once it proves itself as more stable and secure, more will follow. It may look like a small victory, but wars are usually won that way.
Oh yeah, and an obligatory link:
http://www.novell.com/products/portal/
Am I the only one failing to grasp the importance of Web Services?
.NET Web Services are great for certain applications (anything can be overused). I've seen a lot of PHP, ASP/COM, and Cold Fusion code that could have been so much easier to maintain if they would have stuck some of the core logic into a "distributed object model" of some sort.
The concept behind a web service has less to do with "metered services", and more to do with managed code (like EJB's). Internally for my Fortune 1000 corporation, I can share the "login authentication code" amongs 10 seperate applications without the applications needing the compiled Servlet or DLL on the machine itself. Moreover, applications like Microsoft Passport (or similar solutions) can be seamlessly integrated into non-Microsoft sites. Both of these scenarios promote incredible Code Reuse, at the expense of the overhead caused by HTTP and XML serialization.
Both EJB's and
It seems to me that good coders with PHP and apache knowledge seem to be able to pull a better magic than all the Javas and dotNets of this world.
It's really all about the bigger picture. Sure, some little site that uses a couple of include files and global variables can be managed fine by PHP, but it's all about scalability in terms of the applications power and complexity. Now, I'm not saying PHP or CF can't do these things - they can. It just get's to be more of a kludge then a solution as the project requirements get larger ).
There is no longer anything that can be done with computers that is nontrivial and clearly legal. -- Paul Phillips
So it's nothing short of miraculous that Apache managed to retain its market share for about two years while essentially treading water. Let's face it, in spite of a few point releases, Apache hasn't introduced any significant user features in two years.
/these/ are the strengths of Apache.
/any/ job we throw at it. Jakarta is a great example of a web-services enabling extension to the Apache project.
This is a completely unfair statement. While work on a 2.0 release of Apache has been ongoing, development has hardly stalled in the 1.x series. The strength of Apache was never the server itself - It was the modular design. Mod_rewrite, libphp, etc -
Yes, you can run PHP under IIS/Apache for Win - But you have to run it as a CGI. It loses one of it's major strenghs.
Apache is constantly being refined and extended with modules. If the author has any doubts about this, I strongly suggest he 'grep -v ^LoadModule' his httpd.conf. I suspect he would soon realize his Apache did nothing at all.
My guess is the author has never actually admin'd Apache. He's probably been just an Apache user his whole life. (I see nothing to the contrary in his bio at the bottom of the article). Apache is a wonderful tool that is upto
Please excuse the poor thought-line of this post:) I was in a hurry to get my thoughts out the door.
- James
Microsoft is giving up control of C# because C# doesn't matter - it's the free lunch that comes with the $15 soda. The object library underneath C# is where all the value is (and also where the bottleneck is), and that isn't standardized (nor will it be).
For crying out loud, it's even has bindings to a bunch of different languages! C# allows competing languages, but they all use the CLR. Doesn't that show you where the one tollbooth is actually going to be?
- Webservices are simply RPC via XML over HTTP which can be implemented in any language with a sockets library and on any platform with a web server.
- The predominant web services protocols and standards are open and are in fact W3C recommendations or are soon going to be including XML, SOAP, WSDL, and UDDI.
Whenever I read articles like the one referenced in the above post I can't help but feel that people like screaming like Chicken Little simply to hear the sound of their own voices. The fact of the matter is that the goal of web services has always been for interoperability between platforms and languages hence the use of XML and HTTP, heck even Microsoft's Hailstorm claims to be language and platform agnostic with regards to accessing its web services.Repeat after me, all you need to do web services is a web server and a programming language with a socket library and strings support (i.e. almost all of them). Everything else is syntactic sugar and icing on the cake to maximize developer productivity the same way VB and ASP are supposed to versus C++ and Perl CGI.
Yep, it's part of the analysis. Both Namezero and NSI just switched from Solaris over to IIS, and they say that accounts for the lion's share of the movement. When you are hosting a huge amount of servers, then when you change vendors, the waves you make are huge, hence this article on the end of open source as we know it (and I feel fine =3 ).
If god had intended you to be naked, you would have been born that way.