London Stock Exchange floored, again The London Stock Exchange is currently closed thanks to the failure of its trading platform. Technically it's in a "halt phase" meaning orders can be cancelled but new orders cannot be made, nor can automatic trading happen. At 8.03 this morning the market was paused as technical problems became clear, by 8.33am trading was …
Why haven't they tried the tried-and-trusted recipe with Red Hat?
But noooo... they tried to be different - and went for OpenSuSE instead...
Now Linux getting a black mark all because of that decision...
Or is it something wrong they're doing?
Ballmer - because by now he's probably smiling and chortling with glee...
I don't know why people think this is an open source vs closed source thing. The transaction software running the new system is closed source. I'd be very surprised if the OS had anything to do with the outage unless it was a broken driver or similar. More likely it's the application software running on top which has keeled over.
Why the assumption it's an OS issue at all? Seeing as the move to Linux coincided with a move to a new trading system written by Millenium IT, why not blame them? Or the fact they moved from C# to C++ ?
The fact LSE also had problems with their SOLA system, as well as the TradElect (Windows .Net system) they still use in Milan, would point to this not being an OS issue at all.
I hope you don't work in IT with those analytical skills.
Software apps in the stock/clearing industry are a crock of shit regardless of the OS they run on. Blame the developers of such software, not the poor suckers who have NO CHOICE but to use sub-standard products in a live environment.
Maybe using something other than COBOL, Delphi or half-arsed college-kid attempts at Java for would be a good start. You only have to look at founders of such companies to get an idea of their professional merits...
I worked on a Stock-Exchange system and I can tell you well-qualified engineers took their time to do it correctly. It was done in C++ and Cobol on Solaris and VMS, but it could also have been done in Delphi by these people, because they would not fear hacking a kernel or using FPGAs, if really necessary.
Not everybody is as incompetent as LSE.
it does seem to be a fixation - usually of IT managers with a half dozen braincells - that something written in C++ really really badly is intrinsically superior to something written in COBOL really really well.
Being written in Delphi doesn't seem to have dampened WinRAR or Skype or Partition Magic's success, now does it?
The WinRAR, Skype (client) and Partition Magic software are built using the technologies and disciplines appropriate to a single user desktop environment.
I would hope that LSE's MillenniumIT system is built using the technologies and disciplines appropriate to a high volume, mission critical, transaction processing environment - if it isn't then that's the problem
I believe the reason they went to Linux is because it can be a general purpose real time low latency kernel. Which is unique. It the fastest trading floor in the world and others are adopting.
It could be the OS, in conjunction with interrupt driven hardware, that's at fault but the combination is a difficult one. Real time in Linux is a only few years mature. The LSE trading system is much newer than that.
Software is never finished.
I hope this Linux platform goes from strength to strength.
"I believe the reason they went to Linux is because it can be a general purpose real time low latency kernel. Which is unique. It the fastest trading floor in the world and others are adopting.
It could be the OS, in conjunction with interrupt driven hardware, that's at fault but the combination is a difficult one. Real time in Linux is a only few years mature. The LSE trading system is much newer than that.
Software is never finished.
I hope this Linux platform goes from strength to strength."...... HighlightAll Posted Friday 25th February 2011 15:32 GMT
It seems as if Uncle Sam's markets and trading software needs to be finished/erased and replaced with a honest to goodness one, for the increasing suspicions/realisations are that it supports and drivers a Ponzi System ........ http://www.washingtonsblog.com/2011/02/dc-politicians-deal-with-financial.html
what the problem is. Blaming the OS (or even the programming language!) because you happen to know what that is would be a bit like blaming the ball manufacturer for England's failure to win the world cup.
Yes, it *could* be the problem, but the chances are it's something more complicated than that.
When they should probably be laughing at some toss-pot suits called "consultants" who designed and built (but didn't test?) the system. How about an article about the people who built it?
Major world trading system crashing. Ridiculous.
Here's a study for someone: How much does bullshit actually cost the world?
Not just here, but every day in almost every article. What kind of readership do we have these days?
I don't know anything about this problem, but from what I know of linux and software in general, it's most likely to be an issue with the custom trading system software that's been put in place. People use linux as a server because it's stable and highly customisable, and one less thing to worry about in a complex system. It's highly unlikely that it would be an OS issue to have brought about this issue, specially given the redundancy you have in place in distributed software systems.
Companies really need to come to grips with the fact that you can't just surgically remove the beating-heart technology of an organisation, replace it with a new one and expect everything to work as normal. There needs to be buy in from the multitude of non-technical teams that need to help deliver it and most importantly the poor buggers that will need to use it.
The LSE got this really really wrong. There was outcry from all of the major trading vendors and most of the members who connected directly at the aggressive timescales and the fact that the upgrade would cost them each upwards of £1m to implement.
The initial tests were an absolute joke. Nothing worked and internal support was non-existent. Several very large vendors forced the LSE to push the delivery back and schedule more tests.
The business analysts and delivery teams within the LSE remained and carried on with the same mentality as when they were working with the Accenture-built TradElect system - that of an arse-covering bloated bureaucracy. The result of this was a massive disconnect between the developers and users and impossible timescales (it was initially touted that the new exchange be up and running in September 2010) being publicized by teams of pissed off analysts determined to ensure that Millennium took the blame when it all went tits up.
It is a testament to the paranoia within the LSE management that the Turquoise failure was initially blamed on sabotage (turned out to be a contractor screwing up a patch release). Even now, the thing is riddled with bugs and the LSE's reputation amongst it's members is shot to hell.
I hope that others take heed from this rather public failure as a result of not following the basics when running a project to replace a system that is relied upon by many.
God damn MS for constantly changing .NET, they've made so much work worthless.
And all the pro MS folk, how that freaking sharepoint working out. Seems like every time I want to download a government PDF regarding financial records the sharepoint is down cause either it's new or has some new bug. You can say whatever you want, but FTP works better than sharepoint and all the tards running it! Or maybe they are adept and purposely screwing things up?
Just tried to access the LSE site using Firefox...
I'm told the LSE site is infested with malware.
That's going to be fun on Monday morning...
"Diagnostic page for www.londonstockexchange.com/home
Of the 1064 pages we tested on the site over the past 90 days, 363 page(s) resulted in malicious software being downloaded and installed without user consent. The last time Google visited this site was on 2011-02-27, and the last time suspicious content was found on this site was on 2011-02-27."
To quote from elsewhere on the Register a "real time data dissemination issue". Of course it isn't easy to know exactly what this piece of gobbledegook means but it sounds like either a low
level driver problem, a protocol problem or possibly a clock synchronisation problem unless anybody has any further insights. These are not easy things to detect during testing especially if the testing was as badly conducted as some of the posts seem to indicate.
At Uni I used to tell students that "if two people can interpret a (communications) protocol differently, they will" and practical work quickly confirmed this.
As an aside a previous poster referred to the LSE web site, according to Netcraft this runs Linux and is hosted in Italy.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
