London Stock Exchange To Abandon Windows 438
BBCWatcher writes "Computerworld's Steven J. Vaughan-Nichols reports that the London Stock Exchange is abandoning its Microsoft Windows-based trading platform: 'Anyone who was ever fool enough to believe that Microsoft software was good enough to be used for a mission-critical operation had their face slapped this September when the LSE's Windows-based TradElect system brought the market to a standstill for almost an entire day .... Sources at the LSE tell me to this day that the problem was with TradElect ...'"
Re:Seems more big bussiness and goverments.... (Score:2, Informative)
Re:Two years worth of use (Score:5, Informative)
It was not entirely outsourced to Accenture. It was a joint Acenture Microsoft project. Microsoft was involved at all levels of development, testing, deployment, and support.
As one poster to the SJVN article (Bernard) points out, both Microsoft and LSE confirm Microsoft's complete involvement in the project:
Given Microsoft's history of FUD, their habitual use of paid commentors, or even dead people writing letters, I think we should all sit back and enjoy the spin the paid apologists will have to go through to tell everyone how it wasn't MS's fault.
I bet that link on Microsoft's own site telling hoe closely involved they were in the project won't last through the Holiday weekend! Any takers?
From the articles comments (Score:5, Informative)
Yeah I read those too ;) So windows issue or not, it is a Microsoft problem :
to wit; page 4:
"In the development, roll-out, and implementation processes, Microsoft worked closely with the London Stock Exchange to ensure not only that they understood their immediate requirements, but that the solution fitted their long-term business plans as specified in the TRM project."
and
"Robin Paine, Chief Technical Officer at the Exchange, says: âoeThe London Stock Exchange was looking for a responsive partner to engage across all phases of the Technology Roadmap programme. The collaborative approach Microsoft offered made it an ideal choice."
Any more questions on whether Microsoft was "really" involved? Then go do your own research -- there never was any doubt.
NASDAQ going on 5++ yrs. stable on Windows (Score:4, Informative)
"Looking at other exchanges, there are trading platforms that have been able to last 10+ years while scaling quite well." - by Jerky McNaughty (1391) on Friday July 03, @09:40AM (#28571007)
NASDAQ is an example of this, & yes: NASDAQ has maintained the "fabled '5-9's" of 99.999% uptime on Windows Server 2003 + SQLServer 2005 (in failover clusters) since late 2005, acting as the official dissemination system of official trade data:
----
NASDAQ Migrates to SQL Server 2005:
http://windowsfs.com/enews/nasdaq-migrates-to-sql-server-2005 [windowsfs.com]
&/or
NASDAQ Uses SQL Server 2005 - Reducing Costs through Better Data Management:
http://blog.sqlauthority.com/2007/09/17/sqlauthority-news-nasdaq-uses-sql-server-2005-reducing-costs-through-better-data-management/ [sqlauthority.com]
"NASDAQ, the worlds first electronic stock market replaced its aging mainframe computers with Microsoft® SQL Server 2005 on two 4-node clusters to support its Market Data Dissemination System (MDDS). Every trade processed in the NASDAQ marketplace goes through the system with Microsoft® SQL Server 2005 handling some 5,000 transactions per second at market open. The system also responds to about 10,000 queries a day and is able to handle real-time queries against data without slowing the database down."
+
Case Studies - Financial Services:
http://www.microsoft.com/sqlserver/2005/en/us/cs-financial-roi.aspx?pf=true [microsoft.com] [microsoft.com] [microsoft.com] [microsoft.com] [microsoft.com]
"NASDAQ Deploys SQL Server 2005 to Support Real-Time Trade Booking and Queries
NASDAQ, which became the worlds first electronic stock market in 1971, and remains the largest U.S. electronic stock market, is constantly looking for more-efficient ways to serve its members. As the organization prepared to retire its aging large mainframe computers, it deployed Microsoft® SQL Server 2005 on two 4-node clusters to support its Market Data Dissemination System (MDDS). Every trade that is processed in the NASDAQ marketplace goes through the MDDS system, with SQL Server 2005 handling some 5,000 transactions per second at market open. SQL Server 2005 simultaneously handles about 100,000 queries a day, using SQL Server 2005 Snapshot Isolation to support real-time queries against the data without slowing the database. NASDAQ is enjoying a lower total cost of ownership compared to the large mainframe computer system that the SQL Server 2005 deployment has replaced."
----
NOW - the actual PROOF of that "stability/uptime":
http://www.nasdaqtrader.com/Trader.aspx?id=MarketShare [nasdaqtrader.com]
"NASDAQ is renowned for its high performance technology and has proven reliability with 99.999+% uptime. Whats more, firms count on NASDAQ for unsurpassed speed and tested capacity to execute trades quickly and efficiently."
----
AND, "There ya are"... Evidence of the possible stability, security, & speed on Windows, in a high tpm environs, keeping stable & running F A S T 24x7 for 1/2 a decade++ going strong, acting as the official trade data dissemination system for NASDAQ!
APK
P.S.=? Personally, & especially based on the evidences here (the thread topic itself, & the NASDAQ data I just provided here)? Well - I think a great deal of stability & uptime has to do a LOT with the skills of those architecting a system, first, AND later those that have the task of maintaining it also (this means the network engineering staff AND coding teams around said projects), as well as their personal work-ethics - not so much on the Opera
Not at all a troll (Score:1, Informative)
Microsoft promised their operating system was capable of meeting demands. Microsoft was heavily involved in the implementation of the system.
Microsoft knew their operating systems don't do real-time processing, but insisted they could meet the demands anyway.
I seriously doubt this is a "political move". The cost of this transition will be high, and if their current system was salvageable they'd keep it.
Re:Not Windows' fault (Score:5, Informative)
I worked for Accenture in one of the "delivery centres" in the Eastern Europe and it was total crap. They hired 1st and 2nd year students for peanuts, and sold them as professionals to rich foreign companies. The turnover of staff was about a third - after one learned something, it was best to get out of there as soon as possible. From the posts on the glassdoor i can infer that this is the strategy accenture employs worldwide.
accenture just sucks.
Re:Not Windows' fault (Score:5, Informative)
Yes i will second that, responsiveness is one of the most important aspects to traders...
They have special keyboards tailored to their particular trading applications so you can enter trades quicker...
They have dedicated lines between sites because a vpn going over the internet would be slower...
They use unencrypted and often unauthenticated protocols to reduce the overhead.
They intentionally use very short or no passwords so they are quicker to enter...
Security, cost, all secondary factors to the need for low latency.
Re:Not Windows' fault (Score:0, Informative)
No, that was Arthur Andersen. Andersen Consulting was a separate company. They changed their name to Accenture to distance themselves from the failure of Arthur Andersen, because so many people mistakenly thought that Andersen Consulting and Arthur Andersen were the same company.
Re:Not Windows' fault (Score:3, Informative)
Andersen consulting split off from Arthur Andersen. They WERE the same company, and they shared the same rotten management attitude. Accenture is just as sleazy as their former auditing partners ever were.
Re:Not Windows' fault (Score:3, Informative)
As I recall, Microsoft agreed to 'fully-support' the operation. If that meant Dev-Level folks to test & crush London Stock Exchange Bugs, then MS had the responsibility along with Accenture to ensure such a thing didn't happen.
It seems to me that even in a support-role, MS would have been involved directly in the on-going data-center architecture, for example.
Epic-Failure for sure.
Re:Elementary, my near noob (Score:5, Informative)
If I recall correctly, the TradElect platform was pushed by MS as a showcase for its new .NET 1.0 platform, and was written by a whole team (or two) of outsourced developers in this brand new, shiny technology. They ported it to .NET 1.1 but I'm not sure they ever tried to port it to .NET 2.0.
Did I mention it was developed in conjunction with Accenture? Oh, and it was developed by Microsoft itself by all the news reports (though I bet the development was done by cheapest devs Accenture could hire, with a few MS consultants discussing architecture and collecting big fees - it didn't cost $40m in hardware alone!)
So, yes, it could have been poorly written code, but as you say, you can write poor .NET code. It always seemed to me that the project was akin to an 'enterprise java' one of yesteryear - big, slow, over-engineered, poorly developed, resource intensive and generally 'too big to fail'. Seems also that the LSE knows better than to hang on to the worst kind of crappy software and try to make it better.
As for MS bashing, they're the big boy, so they always take the hit. If all software written for MS was great and worked perfectly, we'd worship them as gods. As it is, we continually see problems and give them no slack. If they were a small company trying hard to make a difference, we'd be more forgiving.
Re:Not Windows' fault (Score:2, Informative)
Not arguing your point. Just making a note that Accenture is a Microsoft backed consulting firm. Had a friend who used to work for them, but really worked for Microsoft. Just saying. So yah, it is a Microsoft problem, and is most likely due to just a poor programming team.
Most people have never tried either (Score:2, Informative)
Re:Not Windows' fault (Score:4, Informative)
I have to side with the "accenture is worse than incompetent" crowd.
I know of one project they worked on for the University of Minnesota redoing their financial system that they fucked up completely. I've a friend who was in the periphery of the project (he knew some of the key developers) and saw it all coming. They hire monkeys to produce documentation, and produce complete garbage code. They actually had to fire some people because they discovered they were never at their desk, but produced code. It was discovered they contracted their own jobs out to someone in India to do.
I also know someone who had to work with the "finished" product when it was first roled out, and it was a complete train wreck. (Think magic formulas and tea leaves to get what you need done). It's still largely a train wreck a year later, people have just gotten used to the train wreck.
Re:NASDAQ going on 5++ yrs. stable on Windows (Score:5, Informative)
Clever marketing, but irrelevant. As you note - in passing - this is for the information dissemination system, not for the trading system. It's there so people can do non-real-time-critical look-up of past trades. NASDAQ wouldn't trust their trading system to Windows.
Every trade processed in the NASDAQ marketplace goes through the system
Yes, it goes through the Windows stack after it has been processed by the trading system. Which used to run on a POSIX system on MIPS Tandem hardware [cnet.com] the year after your MDDS system was installed. I can't find anymore recent info even on the NASDAQ site [nasdaqtrader.com].
Re:Not Windows' fault (Score:2, Informative)
From my experience, that is also the pattern in the US. Accenture has repeatedly made the news here for failed projects, yet the governments still turn to them for project work. Accenture must have really good sales people.
Re:Elementary, my near noob (Score:3, Informative)
Actually who planned the entire system that way should get a headsmack, if you need dedicated response times you gotta go with a realtime system, no matter what. You cannot push an os which clearly is not realtime (by Microsofts definition it is soft realtime which means sometimes it reaches it) and on top a vm based system which is not even closely realtime in its garbage collection strategies and then expect to get something useful out of it!
The poor sods ad accenture (I assume as well some outsourced engineers) now will get the smacking but they are probably the ones who said for the beginning this cannot work!
Re:NASDAQ going on 5++ yrs. stable on Windows (Score:5, Informative)
Re:Seems more big bussiness and goverments.... (Score:5, Informative)
...Airbus's belief that humans cannot react as quickly to dangerous situations as computers and Boeing's belief that computers cannot make judgments as well as humans in dangerous situations.
I used to work for Boeing. Boeing's IT department is a huge fan of Microsoft products (Possibly due to the proximity. Boeing Commercial HQ is just a chair's throw away from Microsoft). Engineering's decision to provide pilots absolute authority over autopilot functions stems in part from their experience in dealing with Microsoft office systems.
LSE has horrible functionality and performance (Score:2, Informative)
LSE is not even big comparing to US equity markets. I would even say small. Their data flow is about one tenth of that of Nasdaq. The number of traded issues is a tiny fraction of number of US issues too.
And yet they have absolutely horrible latencies of order entry interfaces comparing to what's 'normal' for US electronic trading venues.
Their TradElect platform even lacks proper timestamps on Level2 market data messages. On top of that wire format description is afair ~700 pages long. Compare it to Nasdaq spec - 12 (used to be 4) pages for market data interface, about 20 for order entry.
It's clear that LSE is one tangled mess - and the reason for it was a monopoly on trading British names and other listed securities. Now with MiFid EU directive in place they do not longer have that cushion.
So do not blame Windows - blame inept management and their boneheaded decisions.
Re:Check YOUR sources (2005), mine (2006 and 2007) (Score:2, Informative)
Sorry you got stuck with this crap job of FUDing on
The system you described probably does in fact exist inside of NASDAQ, although you can't be certain because they make no mention of MDDS whatsoever on their website. Go ahead and search for it at www.nasdaq.com. In fact, search for it on Google as well, you get citations only from microsoft.com, msn.com, and windowsfs.com. Now try searching for SuperMontage, or TIBCO and NASDAQ, you will get the phone book.
You honestly think NASDAQs quote matching system's back-end is a SQL server???
As for your "99.999% uptime" claim, that claim is made by NASDAQ for NASDAQ trading, not for MDDS which they do not even mention or market. So there is no proof that the SQL server cluster trumpeted only by MS, MSN.com, and windowsfs.com has 99.999% uptime. Try again.
Your sources are Microsoft, MSN, and a website/magazine about windows funded by Microsoft. The Iranian national news service couldn't be more biased than that.
Re:Similar at several European banks (Score:3, Informative)
I just noticed I left in the [1]. There was originally a footnote, but it sort of ruined the joke.
[1] Ceci n'est pas une footnote
Re:Did I say it was not for MDDS? Even once?? No.. (Score:3, Informative)
You said I implied something other than this -> NASDAQ has maintained the "fabled '5-9's" of 99.999% uptime on Windows Server 2003 + SQLServer 2005 (in failover clusters) since late 2005, acting as the official dissemination system of official trade data
Exactly. That's what you're implying. That what your links imply. That is not, however, what those links state.
I am not implying anything, I said what I said above in bold, & that's that - you had best prove your accusations of me implying anything.
From the article... 99.999% uptime @ 5k - 10k tpm rates on Windows Server 2003 + SQLServer 2005, acting as the official trade data dissemination system.
Great. Show me the article saying that Windows Server 2003 + SQLServer 2005 with failover clusters accounts for their 99.999% uptime figure. Show me the article saying that the official trade data dissemination system has a 99.999% up time. You have that 5 9's figure. But it doesn't say anything about what system(s) are being considered for the blurb.
Re:NASDAQ does not equal "NaSdAq". (Score:3, Informative)
Once again - a wall of text that says nothing. Here, let me do you a favor and quote your own source:
In an environment when milliseconds matter and best ex is paramount, firms shouldn't trust in anything less that the most reliable trading platform. NASDAQ is renowned for its high performance technology and has proven reliability with 99.999+% uptime. Whatâ(TM)s more, firms count on NASDAQ for unsurpassed speed and tested capacity to execute trades quickly and efficiently.
Now if you can show me where it notes what they're talking about beyond "high performance technology", we might be able to clear this up. I'm especially interested where it mentions Microsoft products. Or even MDSS.
Here's a troublesome detail for you. The closest that source comes to mentioning a specific system is "the most reliable trading platform." And since we've already established that MDSS isn't a trading platform, it doesn't look like this has anything to do with your topic. Of course, who knows? The blurb gives no details as to what they're referring to. What systems? What qualifications? What the heck does that number mean other than a nice little marketing blurb?
You claim that people are attacking you and not your data. Yet you refuse to admit your data is lacking. You've drawn conclusions that have no justification despite all your name-calling and creative formating.
You want to prove your point? Come up with better data.
And as an aside - I don't have to put words in your mouth. You're doing a fine enough job on your own.
Re:trouble? turn off anti-virus, now! (Score:3, Informative)
You don't really need antivirus on Windows unless you are using the machine as a desktop, or the machine serves arbitrary files (i.e. it's a fileserver, email server etc) in which case the AV is there to protect the users, not the server, and would be just as needed on a linux server.
Either:
1: The infection relies on someone receiving an email or visiting a web site, in which case a server shouldn't be affected.
2: Or it is a vulnerability in a public service on the machine, in which case an anti-virus program won't keep the infection out.
Antivirus isn't as necessary as it used to be, I don't remember the last time I had a positive from my virus scanner. I keep some actually infected files around to test it every so often, so it does still work :)
Couple of Quick Points (Score:1, Informative)
First, anyone else notice the obvious issue of this application STILL USING SQL 2000, which reached end of regular maintenance nearly a year ago (Aug 4, 2008)? Seems to me that if this system was structured such that they couldn't easily move to SQL 2005 with Microsoft on the team building this solution, then it may not just be about the stack, but what was done to the technical platform (DBMS, OS, servers, even SAN) to tweak it to reach the performance service levels required. In short, it really is as much about the platform as the application - for this kind of system, you really can't separate the two. As such, a key ownership issue for anyone using MS software as their technical platform is the software lifecycle and maintenance requirement. I can certainly understand why a "2007 go-live" of this system meant they couldn't move to SQL 2005 at that time; but inherent in selecting SQLServer as the DBMS here was the acceptance by the LSE of the MS software cycle - and the need to upgrade in the next 24 months or so. SQL Server 2000 is now an 8 1/2 year old product; even with extended maintenance, I would not view it as suitable for use with something as important as a Stock Exchange. At the very least, I wonder if a combination of Windows Server 2003 and SQL Server 2005 for x64-based servers might be better able to address some of the performance issues here. Finally, I also wonder how much the LSE likes "Patch Tuesdays", especially when Microsoft releases OS and SQL Server (and even .NET) patches with "critical" status - meaning apply immediately. I bet that's lots of fun.
Secondly, I wonder about this being "a system of one". While Linux may not be quite up to the various commercial UNIX platforms in terms of robustness (don't shoot me; this is based on experience, and the gap is shrinking), using Linux and the tools common to both Linux and UNIX platforms - which have a very long (15-25 years) and evolved history together - means there are likely other solutions that are comparable to a Stock Exchange that could be used for "lessons learned" and leveraging for an understanding of how to build this application right. In this case, however, we are talking about a fairly new technical stack - Windows 2003 Server (2004 stability), SQL 2000 DBMS (2002 stability), .NET (how many versions since initial release? 3.5 now?), etc. - being used to create a new ground-up custom design. So choosing the MS stack for this also leads to a "vertical isolation" issue - given the relative newness of the tools, and the limited number of similar examples to draw from, does the choice of a platform here also make it more likely that addressing problems that come up would be more difficult? I know - a "chicken vs egg" conundrum - but this is a real concern, and one of the reasons why MS should share responsibility here with Accenture. Based on reading the 2007 press release, the LSE was apparently to become an example to the world of how the MS stack and commodity (HP) WIntel platforms could do the job that used to require Mainframes, VAXes, Tandems, or big UNIX servers. But as an earlier post has suggested, mainframes still exist for a good reason - 40+ years of experience running these kinds of workloads for specific industries requiring very high transaction volumes. I am not saying the MS technical stack won't eventually be able to run this kind of workload, but you can't shortcut the process of learning what it takes to achieve the performance and robustness required here, at least in my opinion.
For example, using SQL Server insures that you MUST use Windows Server as your OS in any event; Oracle, DB2, or even a custom solution could be placed on a number of different platforms that might have provided the ability to create a "hybrid" solution of Windows/.NET front-ends running against a UNIX/Linux (Oracle RAC?) or even Mainframe/DB2 back-end. So no using "the best of both worlds" either. This means having to learn the hard way how to scale up a SQL Server 2000 DBM