Warning: The Content in this Article May be Inaccurate
Readers have reported that this story contains information that may not be accurate.Sponsored by Travelzoo
$45 & up: Huge Hotel Sale view!
travelzoo.com - Great discounts on rooms for the holidays and beyond!
186 Comments
- inactive, on 09/11/2008, -10/+103Breaking News. Linux enthusiast writes article praising Linux and critical of Microsoft. Next up, Pope admits to being a Catholic, and yes, bears do ***** in the woods.
- Philluminati, on 09/09/2008, -8/+79I enjoy taking the piss out of Microsoft as much as the next guy, but posting this in Linux/Unix does kinda misrepresent the community.
- Stroggoth, on 09/11/2008, -13/+78LOL. The lack of experience on here shows: if you think switching from Win2k3 server to Linux would solve issues like this, you have much to learn. Failures of this scope are design oriented, not operating system or tool oriented. Some of the fastest and largest systems in the world run on .Net.
If a programmer tells you the solution to any problem is to recode on open source platforms, fire them. The first analysis should be what went wrong and how to fix the design, regardless of platform, because in general, they ALL work. - Arowin, on 09/11/2008, -5/+59Sorry, maybe I'm missing something here but the only proof he seemed to have that it was a MS fault was that it ran Windows, .NET and SQL??? I would hazard a guess that perhaps it could have been something else maybe (Not saying it was, just that there wasn't any proof)?
Also, in the first comment on that page:
http://www.computerworlduk.com/management/infrastr ...
"A network software problem was behind Monday’s disastrous system crash at the London Stock Exchange." - Gumboot, on 09/10/2008, -24/+72The Netflix website was down for 12 hours in March and strangely no one blamed Linux... double standard?
- frink, on 09/10/2008, -8/+53SJVN recommends MySQL for a stock exchange! wtf?
- weiran, on 09/11/2008, -3/+43The people commenting here clearly know nothing about writing highly scalable and reliable real-time systems using .NET, and are just Linux fanboys who jump on any negative (and inaccurate) story about Microsoft.
Overhead has nothing to do with it, C# has nothing to do, SQL Server (even 2000) has nothing to do with this problem, they are all capable of creating the right solution, just as much as any other platform. It's down to the design and implementation, not the platform. - thesteef000, on 09/11/2008, -10/+46anyone saying this is .net's fault and would somehow be magically solved using linux is a total TOTAL idiot.
This has nothing to do with the platform. If it were the other way around it would still be a dumb article written by a guy who got hit with the stupid stick.
proceed to digg down..... - uzusan, on 09/09/2008, -28/+64I like c# and .net, but using it for any real time systems, and especially for one of the most high demand systems you could probably have is lunacy. There is simply far too much overhead for that sort of thing. The functions that generate that overhead do come in handy in certain situations, but come on. Pick the right tool for the job.
- cygnus2112, on 09/11/2008, -3/+38http://storefrontbacktalk.com/story/090308netflix
"The IT chief at Netflix has pointed the finger of blame for its site problems last month at "a database corruption event in our shipping system." The problem prevented customers from receiving their DVDs for about three days.
Mike Osier, head of IT Operations at Netflix, wrote in a blog posting on Aug. 22 that it was "a key faulty hardware component" that caused the hiccup. "On Monday, 8/11, our monitors flagged a database corruption event in our shipping system," Osier said. "Over the course of the day, we began experiencing similar problems in peripheral databases until our shipping system went down. It was going to be a long night.""
DB Used: Oracle - PlanR, on 09/11/2008, -7/+41Dont....dont let a computerworld blog ever make it to the front page again.
THX - inactive, on 09/11/2008, -3/+36I write and maintain software for a number of real time control systems (valves and solenoids...) in .net and... Not only are the commenters incorrect, but I got a good chuckle when a guy who wrote the blog uttered the most ignorant and inexperienced word in all of computer software. LOL, he said it!!!! Rewrite! OMG, in real software firms (not computer world apparently) just saying that word in a meeting is grounds for termination. ROFL.
- morphie, on 09/11/2008, -6/+39The reason this was posted in the linux/unix category is beyond me.
And don't blame .NET. And don't say Linux can solve the problem. I mean, would running it in mono be any better? Would using J2EE on windows be better? And what is your profession? Did you write any code for enterprise applications? I doubt it. - Lamtd, on 09/11/2008, -3/+31Indeed, recommending MySQL and Java over .NET and SQL Server for better stability and performance is certainly an interesting point of view... especially when the issue is a software bug.
- jakem1, on 09/11/2008, -2/+29The last time the LSE crashed was in 2000 which predates the installation of this solution. I assume that the solution also predates SQL Server 2005 which explains why they are using 2000. As a result, it's safe to assume that this has worked well for the 3rd largest exchange in the world for at least the last few years. Added to that, we still don't know what caused the crash and it may not have been this software.
This article is fanboy rubbish and anyone digging it needs to take a long hard look at themselves. - moonboots, on 09/11/2008, -1/+25you aren't a stock exchange
- sowdog, on 09/11/2008, -3/+24Aside from "Pshaw! it runs microsoft", can someone explain why it was a bad idea?
- JAgostoni, on 09/11/2008, -2/+23Man ... I rarely comment anymore but really? There is definitive proof that this was a .NET crash? You mean the framework itself crashed ... amazing. I am sure it had nothing to do with poor programming or, as another commenter pointed out, something completely unrelated? I am pro-Linux on servers too ... and I am also pro-Windows ... I have seen Windows servers run extremely high-throughput loads ... and it had more to do with good programming and Architecture .. not JUST the platform.
- MarkBroadhurst, on 09/11/2008, -5/+24I'd love to have an OS and programming language which can keep connected when the network is out.
Can anyone recommend one coze I haven't seen any here. - cRaCKh0rN, on 09/11/2008, -3/+22Digg. A Place for Linux and Mac Fanboys to crap on Windows at any given opportunity. This place has gone sour...
- abigblackman, on 09/11/2008, -0/+19it appears it was a network issue, not related to the servers.
http://www.computerworlduk.com/management/infrastr ... - sjvn, on 09/09/2008, -16/+34It really was an amazingly bad design decision. While clearly I'm not crazy about Microsoft stuff in general, C# is a pretty sharp (sorry) language and .NET can be very useful, but real-time? thousands to tens of thousands of transactions per second? And SQL Server!? SQL Server _2000_ as the back-end!?
Get real.
Steven - MarkBroadhurst, on 09/11/2008, -2/+20I agree the issue has been identified as a network problem (explains "connectivity issues") so regardless of your OS or software it wont make any difference.
- alarion, on 09/11/2008, -2/+20Did the blog author actually suggest using MySQL for a real-time, mission-critical application? I love MySQL, but I would never recommend it for something with this scope. I don't think I could trust it, based on past data corruption issues I have had.
- macko, on 09/11/2008, -4/+21Read this article: http://www.computerworlduk.com/management/infrastr ...
It turns out, the problem had nothing to do with the software. - rprouse, on 09/11/2008, -1/+18Thanks to the others who point out that it is not inherently a .NET problem, but a problem of design. This article was obviously written by a Java fan boy. Suggesting that they use J2EE for systems like this instead? Switching systems isn't the answer, hiring intelligent people to write good software and giving them enough time to do it properly is. You can write bad code in any language.
I am a .NET and a Linux developer. I have worked in Java. I am very pro Open Source, but I'll give Microsoft it's due, the .NET platform is well designed and a pleasure to work with. I liked working in Java, but I must admit, I doubt I am going back.
As for the speed of .NET, the .NET Iron Python is one of the fastest implementations of Python out there, beating even the C implementation. .NET isn't the problem. - darkened, on 09/11/2008, -3/+19ColdFusion? you're out of your freaking mind.
- MtheoryX, on 09/11/2008, -2/+18You would pick PHP over .NET for high-availability, enterprise applications?
- YodaJones, on 09/09/2008, -44/+58They use Microsoft Windows Server 2003 for the stock exchange? What the ***** is in the water over there? Microsoft SQL Server 2000!
Haaaaaaaaa Haaaaaaaa ! - Yazilliclick, on 09/11/2008, -2/+15lol @ php
- brundlefly76, on 09/11/2008, -2/+15Exactly.
Developers of large-scale systems do not rely on the speed differences between languages or databases - the speed and scalability are built into the systems design.
Most developers do not have experience with applications of this scale.
Here is a great example: The Yahoo! user database (which was probably the largest user database at one time) was developed by computer scientists from rdbms companies like Informix and Oracle.
When they designed the Yahoo! database, which rdbms server did they choose? None. It was implemented as a flat-file database. If you understand why, then you are likely familiar with the nuances and specificity of different types of data needs and how they are best addressed on VLS systems.
No one knows enough information about this specific case to say whether is was an error with the database, language, OS, or applications software, but I suspect the latter.
PS I am an open-source programmer. - theeggman, on 09/11/2008, -4/+16Because MS haters have a reputation to maintain.
The simple fact is no one on this board knows what the real problem was or what truly went wrong with the system. Even the author admits that he doesn't know. The article is just an opportunity to push what he prefers. - inactive, on 09/11/2008, -5/+17Um, what's the alternative here, PHP and MySQL?! o.O
- inactive, on 09/11/2008, -3/+14Uh... yeah, pretty much. O.O
- JustLoren, on 09/11/2008, -0/+11I'm reasonably certain you are mentally handicapped.
- Kamujin, on 09/11/2008, -3/+14I love Linux, but when you guys Digg up lame articles just because they bash the competition, you look stupid.
Doubly so when the articles suck as bad as this one. - MtheoryX, on 09/11/2008, -0/+11@brundlefly76:
http://en.wikipedia.org/wiki/Refactoring
They mention it because it's something that people often do before fixing bugs or adding additional functionality.
Refactoring != Rewriting - panic, on 09/11/2008, -5/+15Wow, you really should get a clue...
- ociris, on 09/11/2008, -1/+10I used to be a linux server admin. We hosted thousands of websites on hundreds of servers thats running on RedHat and MySQL. I'll tell you right now.... Linux, PHP and MySQL isn't the second coming. They are very capable, but don't think that its much better than windows and .Net. Both are very capable as long as the programmer knows what they are doing. To say otherwise is either lack of knowledge or just plain ignorance (@smotpoker).
- darkened, on 09/11/2008, -3/+12The .Net development platform is the premier platform for enterprise development now. Welcome to the 21st century.
- jiqiren, on 09/11/2008, -1/+10http://nelsonhaha.com
- shadowspawn, on 09/11/2008, -1/+10From what I've done with benchmarks, .net seems to be a good part of a complete system. The article also said that quite a bit was done in C#.
I'm going to guess high-priority dll's that hook directly into MS for the sub 10ms times on transactions.
Take into account MS's handling of clustering services, that is a stable system. All it really takes, however, is someone to not understand timing when opening and closing connections. I'm going to guess that something was using some type of hardware SQL-based round-robin and a transaction process got ganked or was timed wrong because of a network issue. Lag, basically.
When that happens it begins to snowball quicker than you can blink and when dealing with finances it's sometimes best to press the big red button and close all connections and keep them closed till you figure out what went wrong.
Yep, I think someone pressed the panic button on purpose before people got sued and money was lost or gained incorrectly.
I've seen that happen on linux systems. I have not, however, seen that happen on those that still use Tandem or other mainframes.
I think the reason that people go to MS is for the support. Yes it might not be the fastest, but there are some big numbers that go into production... face it, MS isn't going anywhere anytime soon and it *is* standardized for framework programming.
Also keep in mind a project like this isn't done overnight. Look what was happening in the *nix community when this project was incepted into design phase. There weren't really too many choices, MySQL didn't even support stored procedures so go-between proxy servers had to be used.
Gotta think about what you are spewing. - weiran, on 09/11/2008, -7/+16I think the comments here show how narrow-minded, short-sighted, and Slashdotted the majority of people on Digg are now.
Dicks. - HookmasterCH47, on 09/11/2008, -15/+23So the London Stock Exchange has experienced the Wow?
- aamer, on 09/11/2008, -0/+8I'm glad we have some people of reason in this thread who realize the issue is not inherent to .NET or Linux or anything else, but likely resulted from poor architecture.
Like most people here, I love working with open source, but also realize that Microsoft server products are not usually the cause of a failure. In fact, even if it was running on a mildly ***** platform (keyword: mildly), a good fault tolerant architecture could withstand it. I've worked with 100% Microsoft applications running on over 60,000 servers and can personally say from experience that good architecture is the key solution for/cause of a problem.
The fanboy nature of this article just gets me. I mean, comon, the guy thinks that running the system on "RHEL (Red Hat Enterprise Linux), JBoss, and MySQL" would have somehow been a more solid system? Ok, buddy, why don't you implement your stock exchange's realtime backend on MySQL and get back to me on that. I've worked on enterprise systems that use all of these components, and I have no idea what makes him think they would have been more solid. - inactive, on 09/11/2008, -3/+11Er, no I would say computerworld has failed, they should fire all their bloggers and get people who... um... know some thing?
- MarkBroadhurst, on 09/11/2008, -1/+8It is what was available at the time they started.
If I could have installed SQL 2008 in 2000 I would have. - nyabutid, on 09/11/2008, -2/+9Computing and IT is turning into a cult. We all seem to have VERY deeply rooted beliefs half of which we can justify. This is not the first case of some random douche bag being a whistle blower to a problem they don't have a f'ing clue on.
- hsd31, on 09/11/2008, -4/+11Inaccurate article. Read what actually happened here: http://www.computerworlduk.com/management/infrastr ...
Just some linux fanboy ranting. - gronya, on 09/11/2008, -0/+6Three days on there is no satisfactory explanation. Maybe it wasn't .Net. Possibly the interface between the Microsoft based systems and the network. This article slams the exchange for being in denial and highlights some of what was going on before the debacle.
On thing is clear, the failure to explain what really happened is bad for the London Stock Exchange and it is an insult to hard working IT professionals who want to learn the lessons...
http://www.computerworlduk.com/management/it-busin ... -
Show 51 - 100 of 192 discussions



What is Digg?