Open Source vs. Shared Source

We got a pretty lively discussion going in scripting about this, so moved it here so we can get our techno religions awn.

Open source for innovation and all-eyes review (linux), or managed/shared source for stability and referenceable code state (windows)? Flightless waterfowl or flying Windows?

Innovation RULEZ!!

Go penguins! Go penguins! Is your birthday!
::does little dance to go with that::

There are places for both.

I like the concept of open source, but I have yet to see an open source project that has the polish of a closed source one.

OK, I'll bite. Linux has produced a more stable and secure server platform than windows, in a shorter span of time. Apache is also more stable and secure and by far more widely used than IIS.

Windows has produced a more feature rich desktop, but it is also bloated for that same reason. Same goes for Office vs Open Office. Linux on the other hand offers a dozen user interfacees that go from slim and fast to feature rich and bloated, so the user can choose, based on his/her likes, and the speed of their machine.

Internet Explorer was far better than Netscape for a while, but the decision to open-source Netscape has finally paid off. I'd much rather use Mozilla, Opera, or a new variation on the KDE web brwoser known as Safari (I'm using it now) than IE.

While Microsoft can continue to add features to Windows, Office and IE, the fact is that these are all mature technologies. There just isn't all that much more to BE added at this point. So the open source alternatives have caught up.

As with the story of the tourtois and the hare, slow but steady wins the race.

Right now the only thing I boot Windows to run is Second Life.

PS: I assume you are aware that the SL servers run Linux.

I have had a dual boot windows and linux OS for about 7 years now (since my junior year of HS - 1996). I download new ISOs on a regular basis and have used and set up linux (SuSe, Redhat, Mandrake, and many others that I tried but didn't keep). I have also run an apache web server, and numerous other linux software packages, including KDE, Gnome, Enlightenment, OpenOffice etc. etc. etc.

I never said they weren't more feature rich or more stable.

I said more polished.

All the linux software products I have tried seem to have more rough edges then commercially developed software. And by this I mean that the more polished commerical software is more polished than the more polished OSS. I have seen god aweful closed source software of course, and I'm not talking about that.

I think it comes down to interface design. But I could be wrong. Anyways its really just an opinion statement.

Don't Forget to mention The MacOS.

Apple continues to be a drving force in the PC market, PC Culture and even Popular Culture.

Ah, the guantlet is down! I challenge the statement that Linux is more stable. I say Windows is by far more stable than any available distrib of Linux.

How can I say such heinous things! Well, don't take my word for it. Here's a sample of my data; keep in mind that before Win2K these numbers used to be very, very different, and were frequently used against anybody building a case for Windows:

http://www.tpc.org/
This site is a third party industry benchmark for performance and stability of high end transactional systems on all OSs. Windows pretty much owns every category. Linux is nowhere to be found.

http://www.middleware-company.com/j2eedotnetbench/
A recent benchmark test peformed by the middleware company, an industry icon for J2EE/NonMS development, recently pit the best J2EE implementation against the best .Net. To summarize, even the competition wanted the database run on Windows because the DB numbers and performance were better (far better) than on Linux or Unix. in no category did MS tech fail to outperform J2EE/non MS tech by a factor of two- three, and often more.

Lastly, every year there is a hackathon, it's published on the web. Linux and Apache are the first to fall. EVERY time (along with Oracle). Frequently Windows is never cracked, expecially after Win2K.

Unix has a niche for graphics servers, hardware-specific processing, etc., I'll give it that. For scientific number crunching it's still considered a standard, but that mostly has to do with accepted and tried practices, and available apps, on those OSs. Windows is slowly but surely building that credibility. Whether or not it will prove itself superior at this kind of processing, I don't know.

I'm not a pure MS advocate; when MS isn't right I don't put the tech in a proposal. But in the field when I hear people say "OMG Linux is so much better, more stable, MS sucks!", and I'm looking at doing a job for that org, I challenge the statement with a request for data...ANY data. Never ever has anybody given me ANY convincing data to prove that Linux is more stable than Windows.

Just my bit.

EDIT: oh, one other thing. Netscape was originally an open source project, the first really big and visible one. The results are the current state of Netscape.

Um huh?! Netscape was not an open source project originally.....

It was built by the Netscape company which was later bought by AOL, which later opened it to Open Source. And opening it up I do not believe helped it in any way.

I have used both IIS and Apache for web serving, Windows and Linux. I have -not- used .net for web hosting so I don't know about it, but you are talking about Win2k in your article.

I can say for sure, 100% that the way windows handles permissions is an absolute royal pain unless you have complete control over the box. Linux and apache on the other hand actually allow me to have and set the access controls I need for the specific files I need. I can not do that in my hosted Win space. It has made running a bulliten board system a nightmare. Part of it may be that the person who owns the box doesn't wanna open it up, but they shouldn't have to!

As I said, I have not used any .net framework for web hosting. I have friends who swear by it and I will try it sometime. I also admit that my problem could partially be the fault of the ISP, however I need my freakin chmod. Or some equivalant.

I would be interested in your hackathon. Mostly because windows vulnerabilities are discovered on a weekly basis or near enough.

As for stability.....from experience I would argue that Linux is more stable than Win2k, but not XP. In a pure server only environment. As soon as I start trying to run KDE and Gnome stability rating drops. Or any other software. That is part of the 'polish' I mentioned early that is lacking.

Great stuff Ama. I would submit that your Windows guy is not entirely informed as to best practices for policy and role regarding Windows security. If you approach a windows box with a Unix or Linux framework of security in mind and try to replicate it, you will be very, very frustrated.

Every operating system is constantly plagued by reports of security violations - don't forget it's A-Patchy web server. A big difference is that MS is a target; they're the top of the heap and everybody hates the king of the hill. The Apache bug list and security flaw list is a long, long one. Under this kind of negative scrutiny it's amazing that MS still has the credibility to get anybody to buy it, but the numbers prove the point.

Also, Win2k != .Net...same for XP. .Net is a runtime distributable with a class library, similar to the concept of the JDK. None of the implementations in the TPC-C have run anything .Net until recently. When they do Linux and Unix will entirely dissappear. .Net server 2k3 (and beyond) will have the .Net framework integrated. But you don't have to use it, you can go all Win32 API or COM (shudder).

There is a distinction; hardware specific processing is not what Windows does, and that's what you need for games like SL (I'm guessing this game runs on Unix on the server side). I'm sure Turbine (Asheron's Call) probably uses Unix too. I'm talking web serving and entreprise computing, like financial transaction systems etc.

And btw you're right Ama, Netscape was not initially open source, but the code was publicly available before the AOL buyout. I know because I had when I was filling an R&D spot at a pretty big company you may never leave home without. You're also right about the fact that opening it up to the community didn't do it any good at all.

Just one slight clerification/question ... every report I have heard puts Apache as the most used web server with a fair margin over IIS/MS.

Also my forums I try and host on a Win2k box (generously donated to us :/) drive me insane. It is a php forum, and it runs fine once installed. I just can't upgrade it, patch, etc. Anytime I try the person who works at the ISP and got us the space says oh look its easy and does the exact same things I did. The difference? They are sitting at the box. I have no way of sitting there, and I have no access to setting any permissions at all within that space. Which is something I need on webspace I maintain. I need to be able to decide which folders can be accessed and which can not. I really wish I knew the way you could do this in the environment I am on - subleasing a space on another box. The won't give me admin rights ( correctly so) and I can't remote desktop in. My only access options seem to be through a web browser and through ftp.

The difference I find in linux is that with the same two tools I have much greater control over permissions and thus content.

As I said before I like the concept of Open source.

But here is the key: Each peice of softwre needs to be evaulated on its own.

Whether or not it is open source is a factor in any cost/benefit analysis, sure. But its just one factor. Some OSS software has less support, but more potential features...which is a trade I may be willing to make if I have the time and know how to make up for the lack of support. Etc. etc. etc.


And if we are into MS bashing.... Their products are decent and I don't really fault them. What I don't like are two things: Bill Gates (he's an asshole) and their "Business Practices". You know the only two products MS makes any profit from at all (out of all their divisions) is Office and Windows. And from just those two products they are able to support all their other financialy failing products and divisions, and have lots to spare. Microsoft would break even selling each copy of Windows and Office at about $45 per copy of windows and office. Anything above that is pure profit, paying for their other ventures. And when you consider the cheapest version of windows, which is only an upgrade, is $100 and prices go quickly into the $400+ range .... you start to see where the monopoly is and how it effects things. MS is not dealing in a fair market environment with windows or office, and they are taking advantage of it. And it shows, because they can't turn a profit in any industry they don't have a monopoly in. :/

Well...this is a technical conversation. Business practices are different matter. I agree that I often question the business practice and sell-at-any-cost position of MS.

MS also makes profit off of SQL Server, Sharepoint Server, and a few other products btw. But yes Office and Windows are the cash cows.

Microsoft expends a lot of effort to win these things, and when they don't win they spend a lot of money challenging them.
Forty Billion dollars can buy a lot of things, and I don't have any problem with Microsoft spending their money on winning performance benchmarks. The TPC by the way is primarily about PERFORMANCE and not stability. Your URL references were both non sequiturs to your statement about stability.

Even when it comes to performance though, the TPC benchmarks are no magic bullet. They are used in the real world more as a comparison of hardware than software. In the real world, as you scale up a system based on any of these technologies you have to roll in your own mix of techniques, be it clustering or splitting up SQL and web components. The TPC tests don't generally take these things into account. They look good on HP and Dell glossy literature though

Article at TPC on the pitfalls of relying on default/automated performance numbers:
http://www.tpc.org/information/other/articles/NoAnalysis.asp

Back to stability and security (which pair well together as opposed to pairing either of them with performance), the most widely quoted study that has Microsoft comming out ahead of all the others in the field is this one (which involves a free sign-up to get):

Aberdeen study:
http://www.cheetahmail.com/c/s/edit.cgi?aid=73417168&AFF=AFFhttp://www.aberdeen.com/2001/research/11020005.asp

However is has been widely disputed. Aberdeen's largest customer is Microsoft. They tend to say things that will keep that money flowing.

Here is one of the articles that points out flaws in their results:

Debunking of Aberdeen results (See last article on page):
http://fscavo.blogspot.com/2003_01_01_fscavo_archive.html

As far as how Microsoft is spending its warchest, they had better find the formula for success soon. I anticipate that they will become just another software company in the next few years. They are trying to head off that fate, but so far with little to show for it.

Article on future of Microsoft stock:
http://story.news.yahoo.com/news?tmpl=story&ncid=581&e=1&cid=581&u=/nm/20030420/tc_nm/tech_microsoft_report_dc

The Barron's article requires a subscription....but since I have one, I took a look. Its a bit more detailed and up to date, but the gist of it is contained in an older web page by the author, financial analyst Bill Parish...

Older Article on which the Barron's article is based:
http://www.billparish.com/msftfraudfacts.html

You can't get an unbiased view by trying to find an unbiased source. There is no such thing. I read a lot. And I've made money using IBM, Microsoft, and Unix products on Mainframes, Minicomputers and PCs. The best way to get the big picture is to view the issues from many different angles. Here are some recent sources on security and stability (which I find far more usefull than TPC results):

Internet Security group funded by Department of Defense:
http://www.cert.org/

Current Anchordesk article on Windows stability:
http://www.zdnet.com/anchordesk/stories/story/0,10738,2913409,00.html

Internet Storm Center. Vendor neutral reporting on virus and worm spread:
http://isc.incidents.org/

Security from consultants point of view (primarily a MS consultant):
http://www.securityfocus.com/columnists/127

Newsletter for Windows security issues:
http://www.ntbugtraq.com/

And finally, when I think of the recent hacking (in the bad sense) success stories the names that come to mind are: SQL Slammer, Code Red (I and II), nimda. SQL slammer was (and still is) a disaster for Microsoft. They botched the patch, they mishandled the PR on it, and sadly, it had as much impact on the internet as a whole as it did on SQL Server installations. Many companies mistakenly ignored it thinking that if they were not running SQL Server they had nothing to worry about. Of course they spent the next few days reprogramming their routers (or getting their up-level ISPs to do so).

Can anyone think of an example of any such hack-attack having an impact on a mainframe system?

And yes, to whoever posed the issue, the SL server are Linux, not just any old Unix. I didn't know that until I recently read the posting about the upcoming server move. Maybe some people here think they should be running Windows.

I think AOL had it for a while before they opened it up too. There are mixed views from AOL insiders about the results of this move. As best I can reconstruct it, when AOL acquired Netscape, it was seriously behind Internet Explorer in features, speed, and stability. AOL management wanted that fixed NOW, and the inhouse developers knew that they were nowhere near ready for a release.

Going Open Source might well have been a way to stall for time with upper management. They were certainly disappointed that making the source code available to the public didn't produce Netscape 7 overnight. But also from what I have read, they are fairly happy with the results. Open Source does NOT produce code on a deadline basis. If will always lag behind a fully funded internal organization in that regard. Open Source CAN (not always DOES) produce a stable, working product. Eventually it also produces something with the most requested bells and whistles too.

I agree, both Gnome and KDE have "issues". I've been using KDE 3 lately though and have had few problems. In most cases the problems don't cause KDE to crash, and I can't think of a time when it cause the whole system to crash or lock up beyond recovery.

Windows (as mandated by Bill himself) does as much as possible to cover up the fact that the GUI is built on top of a character mode operating system. So, other than holding down Function keys when you boot, there is no way to recover from anything seriously wrong with the GUI. On Linux systems the GUI is more like an application. KDE and Gnome compete with one another and eventually both benefit from the competition. I'd love to see Microsoft field a "Windows for Linux" GUI to run on top of Linux. I'm not sure I'd PAY much to run it, but I'm sure some people would. I think ultimately this may be one way that MS can get out of the corner they have painted themselves into. Time will tell.

I have known of production Windows servers crashing due to bugs in video drivers. I've also known of production Windows servers operating at half capacity because the playful administrators insisted on running the OpenGL Pipes screen saver. When informed that this was a bad idea, they switched to an old pre-NT screensaver that it turns out had both performance issues and a convenient back-door for hackers. (These were MCSE certified people by the way). There are both benefits and penalties to pay when you set out to produce an operating system that even an idiot can use.

I know some very competent MCSEs too. Most of them have experience with other operating systems too and don't approach their jobs with vendor bias. The "Windows Rulez" types make me very nervous.

Absolutely true!

I love the Mac OS X interface (and not just because of the name).

I'd never used an Apple computer before OS X. I was initially not impressed with the reliability of the system. In fact I switched my iBook to Linux for the past 6 months or so and it didn't crash once. I missed the ability to view DVDs on it though (and that is more a licensing issue than a technology issue).

I've been keeping up with the Apple support forums ever since I ditched OS X and from what I read it looks like they have cleaned it up quite a bit as of 10.2.5, so I am giving it another try. So far so good. I expect by 10.3 this is going to be a VERY nice operating system. As it is I have full operability between my OS X laptop and my Linux machines. I can open up windows from one machine and control them one an app by app basis remotely. This capability is Beta at the moment, but I expect by 10.3 it will be standard issue. The big befit of this is that all 10 billion open source programs written for the Unix environment will be trivial to port to the Apple. This will not only benefit home users but catapult them way ahead in the server market. Exciting times ahead for Apple if they don't screw up.

Sorry Mac, I disagree with your assessment of TPC perf/stability. TPC-C would not post a benchmark of this nature unless the system was found to be stable. Posting high end transaction numbers for unstable operating systems would make TPC-C benchmarks useless and they would have zero credibility. This is by no means a non-sequitor. It is also very interesting that the Unix side now discredits these numbers when four years ago they swore by them.

Let's not forget Sun's statement that "cosmic rays" are to blame for the disasters they found in their hardware...which runs Unix. That is published info and still incredible. If you can't blame yourself, well blame the universe. IBM, and Sun, both put in their licensing agreements that bugs and flaws in the technology are not to be publicized. If you've ever been in a large-scale tech contract with either of these orgs you know what I mean. You can literally be sued.

Oracle did not pull it's "one million dollars" campaign for no reason. It was found to be utter bunk as the database was hacked three days after they published the campaign. The response from Oracle was laughable. Slammer hurt the SQL story, that is true. But SQL is by no means the first database to present serious challenges to its users.

I also notice that your data shows no industry standard positive benchmarks for Linux against other parallel implementations of technology. What they do mostly provide is reasons not to believe the existing data, which proves nothing regarding stability or perf.

Mainframes are interesting, you have a good point. But mainframes are never put as the frontline for access to data from external sources in any well architected system. The primary attack would be to intrude the gateways, which once done, can lead to the mainframe, but only if the system was very poorly designed or the hackers had some intimate knowledge. You don't find a mainframe box in a DMZ fighting off denial of service attacks.

Lastly, I'm not sure why you would indicate somebody stating a preference to run SL on windows...the above posts state that Windows is not ready for this kind of processing/serving based on their preference for hardware abstraction. I'm willing to bet that no competitive analysis of Windows technology, with an actualy sim implementation written for Windows or .Net, was conducted. The reasons they didn't go MS are probably business/existing skillsets. Don't forget that the officer tier of Lindenlabs reads like a last gen executive list of Real.

This is great discussion

Tcoz:

Doesn't look like either of us will convince the other. So I give up after this post (you can have the last word).

I don't think the TPC numbers are bogus, nor meaningless. By their own admission they are not everything (thats why I included the link that said so).

Using TPC benchmarks to select an operating system, or a hardware platform is like using drag race results to select the family car. Business people are more interested in TCO (Total Cost of Ownership) than TPC numbers. TCO can be a very subjective thing. Different businesses have different needs.

You are correct that the TPC number are always promoted by those who win, and put down by those who don't. No surprise there. I don't work for any of these companies, but I have worked for companies that review this kind of stuff and I assure you they don't make the decision based on the TPC benchmarks.

The notion that Microsoft will dominate the server market crested about 3 years ago. I think it is bad enough that they dominate the desktop market, but hopefully that will change too. I fail to understand the logic of those who stand on the sidelines and root for one vendor to take everything over. That's the worst thing that can happen to this industry, and in fact it has contributed to the sorry state that it is in now.

My own *prediction* which just happens to correspond with my desire, is that Microsoft will slowly lose it's grip on the desktop market. They will continue to be a contender in the server market. They will look for alternate ways to boost their stock price. Maybe they will succeed, maybe not. It is not a religious argument for me that Microsoft is evil and other companies are good, nor should the reverse be for you. It is just common sense that competition is good for the consumer. I'd be just as unhappy if all inkjet printers came from HP or all digital cameras came from Nikon.

Microsoft transitioned a long time ago from being a technology company to something out of a bad SCI-FI movie. For them to recover will probably involve a few people stepping down...

http://www.theregister.co.uk/content/archive/20840.html

Lol, bad sci - fi company. I have no idea what that means but I like the sound of it ;p. To MS's credit they didn't get 40bill becuase nobody sees any value to the tech or finds it useful. Once upon a time they were the underdog and nobody thought they'd last. If and when Linux rules it'll be the same story.

I'm with you regarding religion, and fault MS for that a lot. On the other hand, I fault the Linux community for providing data to bash instead of data to prove. Show me a large scale third party apples to apples test where Linux outperforms windows and then I can help the internal Linux guy argue the point when setting up my proposals; "Linux is free" doesn't stand up when you add 40k for the J2EE app server to run on it. And I do provide other data besides TPC-C; I again agree you would be foolish to base a platform decision solely on those numbers. But it IS data generated from use of the tech.