When open-source eats itself, we win
For years the headlines have been about open source cannibalising proprietary software. But what happens when open source starts to cannibalise itself? In some markets, open source rules the roost. For example, Drupal, Joomla, my old company Alfresco and other open-source content management systems regularly duke it out for …
This topic is closed for new posts.
JBoss vs Tomcat?
In application servers, JBoss and Tomcat spar
Both are virtually developed by Red Hat. JBoss uses tomcat as webengine: I know both of the them on source level, here is a ref: http://www.jboss.org/jbossweb
Tomcat is also not an application server.
Re: JBoss vs Tomcat?
@stanimir
"Both are virtually developed by Red Hat. JBoss uses tomcat as webengine: I know both of the them on source level, here is a ref: http://www.jboss.org/jbossweb"
Correct, Tomcat is a JSP engine whereas JBoss does the other J2EE stuff and uses Tomcat for JSP serving.
"Tomcat is also not an application server."
- It's a bit of a semantic question. You can run tomcat as a standalone web server and have it host entire end-to-end applications. OK, they won't be J2EE apps, but they're still apps. Tomcat can host Spring apps, for example.
So Tomcat is an application server, but it lacks features of a full blown J2EE server that has container-managed transactions, security, message queues, beans and other gubbins. But Spring or other frameworks can do that in any case.
Re: JBoss vs Tomcat?
@Eadon,
Tomcat includes the real web server, not only "jasper" that's the JSP engine. We use (heavily) hacked version of the NIO connector w/ socket hijack, modified web-app classloader and some other goodness . Connectors are part of the coyote subproject, still tomcat. "Catalina" is the one that implements the javax.servlet API and still part of the tomcat.
Yes, tomcat does include deploy/undeploy enabled classloader since version 3.2.x, so in a way it can run stuff; it just does not feature full JEE (they renamed J2EE which is weird but "J2EE" is obsolete now).
As for security Tomcat actually supports jaas. Also supoprts JMX registrations. Yet, it's nowhere near jboss when it comes as application server - for instance it doesn't support JCA.
I excuse for the mass topic derail, yet I can't see how the author decided to put "tomcat vs jboss".
Ans since you went for details, I present some ;)
Apache MS scrap
For a decade I've been following the web server battles. Apache was twice as popular as MS and the gap was growing. Then MS started to game the systems, with maneuvers regarding Big Daddy hosters and the like and the gap (at least in terms of netcraft stats closed rapidly. But things returned to normal again (according to the netcraft stats.
Along comes nginx and this seems to be taking share from both players. This is the beauty of diversification. And open source is excellent at diversification.
The decline of IIS suggests that Windows is not trusted these days, relative to previous times, to run on servers connected to the internet in the DMZ - windows is not trusted for serving web pages. This maybe due to poor performance (ASP sites are always sluggish in my experience) or it could be due to the poor security that plagues windows. Having to run - and rely on - anti-virus on servers is a nightmare, and that's what you have to do if you are crazy enough to use Windows as your web server.
Re: Apache MS scrap
YEH! Don't forget Windows! Take that Windows...take that!
You'll learn'm Eadon, you'll learn'm!
Re: Apache MS scrap
Err... The decline of IIS suggests that IIS is not popular as a web server rather than Windows is not popular as an OS. You do know that there are many installations of Apache on Windows, don't you?
Re: Apache MS scrap
How often would you run either nginx or Apache on Windows? That is a good question. Both apache & nginx configuration have a *nix mentality. Making one leap away from the Windows IIS gets you half way to a *nix system.
Like Eadon has correctly mentioned, there you wouldn't waste your time on crap like antivirus
Re: Apache MS scrap
I'm certain any Windows sysadmin worth employing would be capable of making quite a secure Windows box for serving web pages. Windows has improved, as has IIS.
That said, there are a lot that run around with "MCSE" tickets that barely know their UDP from their UTP. There are some with "RHCE" tickets who are just as bad. I'd say in both cases, this is a case of the odd few bad apples, not everyone with these qualifications is this bad.
Windows will always have a bigger footprint than what can be achieved with Linux or BSD, granted, but the difference isn't that great, and very few systems would run such a bare bones configuration.
As for what I choose? I use Linux largely because it's what I'm familiar with. Not necessarily because it's better, but because it's what I understand better. I know the common gotchas for security and can set up a machine to resist a fairly wide gamut of attack vectors.
But I won't catch everything, this is where vigilence and careful monitoring come in. Antivirus and intrusion detection systems are tools for exactly that sort of monitoring.
The other tool is regular backups, so you can recover when disaster does strike, as not every failure can be blamed on miscreants on the world wild web!.
Who's writing your news?
I can't help to think that this article was written by someone who just doesn't understand software progresses, open or closed source. If we don't strive for something better, regardless of reasons, why do anything at all?
I'm not trying to flame the author of the article, but maybe the author jumped the gun by using the concept of eating ones self. After all, is progression eating ones self, or bettering ones self? So maybe the tone of the article should of been presented as "Open source just keeps getting better."
Also, if you start counting which open source domains are getting better and better...you'll run out of ink (or memory in this case).
Nginx
I switched all my VPSs from Apache to Nginx a few years back because... well, if you must know... because Nginx had a really cool looking soviet style red star as an icon, whereas Apache had a poofy feather.
Well, it's as good a reason as any other I've heard.
Re: Nginx
This is by far the best reg comment i have ever read. I just burst out laughing right after it! Great reason!
Title click bait
I mostly agree with the author. However the title seems to indicate FOSS is killing itself off.
The good thing about diversity in open source software is...
...that it doesn't come with the usual drawbacks. If you have 2 competing projects, chances are the teams are going to talk to each other and even share code and ideas.
They may disagree one some points, but they are free to agree on others.
Let's take a negative example of diversity, ARM SoCs. There are literally dozens of different SoCs out there and they are all different. Partly since they were designed for different markets (a router doesn't need graphics), but mostly to keep developers from switching to another SoC.
This is why, when you are an operating system kernel and find yourself running on an ARM SoC, you have no idea of how your system looks like. You don't know where and how large your RAM is, you don't know how to access the serial port or where it is in IO space. You are completely blind and can neither discover your hardware, nor blindly assume it's at well known addresses. That's why most resources of open source projects running on ARM currently go into supporting all the wildly different hardware platforms. While on a PC you might cope with not having accelerated graphics or a non-working WLAN chip, on ARM have to cope with non-working flash or no frame buffer access at all.
That is negative diversity: Trying to be different to make life hard for people.
Re: The good thing about diversity in open source software is...
This is only negative from a abstraction PoV. The ad-hoc mix&mash of electrical signals and peripherals to the SoC is what makes your router cheap.
You confuse purpose built software for general purpose full blown OS distributions.
Re: The good thing about diversity in open source software is...
It's also the SOP of a chip company. The jargon phrase (well, it was back in the 80's when I was with Intel (name-drop)): "Own the socket"
i.e. once the design decision has been made for YOUR chip, then every sale of the end-product is a sale for you, and not one that any of your competition will get.
One of the reasons open hardware is slow to take off.
dodgy data and misleading stats again.
Nginx is most often used as a reverse proxy, not a web server in its own right. So whilst you'd have a an apache with mod_proxy infront of dozens of apache / tomcat / etc instances, it's often nginx being the one handing of the the apache instances behind to actually DO the work.
But the HTTP headers will say nginx as that's what the client is actually talking to.
Apache and Nginx work well together, not necessarily one vs the other.
Re: dodgy data and misleading stats again.
Indeed. This is exactly how we use it. Though we strip the HTTP header.
That's simply a blind point of view. Open source is EVOLVING more rapidly than proprietary software. That's because better technology has no licensing or commercial barriers in substituting obsolete technology.
And also you choose a very limited example: web servers. Look at what happened to version control systems in software development!!! I would choose this field of application, if I were to write an article about open source "eating itself"!!!
CVS was completely cannibalized by Subversion, which is being played down by Git. Is that really cannibalizing? NO! It's evolution, because the version control systems in the meanwhile gained tremendous features.
The proprietary world in comparison is slow and lacks of evolution (cannibalization). Even Microsoft had to offer Git as an option to its developers...
What a surprise - Not
Every software project starts small and grows features; the more users it has, the faster it improves, and the more it improves, the more users it gets, in a kind of positive feedback loop. (Every software project eventually ends up incorporating a Turing-complete programming language, as well, whether intentional or otherwise.)
Sooner or later, the options you can pass to configure grow past a screenful, so everyone just builds it with everything enabled whether or not they need it. At this point, it acquires a reputation -- whether deserved or not -- for "bloat".
Sometimes, someone forks it and chops out chunks (Firefox was Mozilla minus lots of stuff). Other times, a whole 'nother project comes along, started from scratch, and displaces it. Until it, too, starts growing features ..... And the whole cycle starts over again.
Everybody wins?
The graph shows a *loss* of 0.01 percent for Apache and NGiNX combined, for the graph 'Market Share for Top Servers Across the Million Busiest Sites'. So, Apache/NGiNX combo - the open source part of the graph - lost 100 of the most busy sites on the net. It might not amount to much, but how does everybody win - everybody being open source according to the article - with diminishing open source in the arena?
Signal to noise here...
Seriously, why are there few up votes on the three comments saying nginx is used as a reverse proxy and the backend is Apache? They are not eating each others lunch as much as the author purports...
We moved a client to nginx some 5 years ago as a RP for hosting. Between splitting media content to different host names, moving static content to minimal nginx backends, and using nginx in front, the popular sites don't see the anywhere near the performance issues at the busy times. It's not rocket science.
(The backend load balancing is another matter completely - urgh!)
This topic is closed for new posts.
