London Stock Exchange Finishes Switch To Linux 106
DMandPenfold writes "The London Stock Exchange has successfully set into live trading a new matching engine based on Novell SUSE Linux technology, following successful last-step setup procedures on Saturday. The move has been billed as one of the LSE's most significant technological developments since the increasing prevalence of electronic trading led to the closure of the traditional exchange floor in 1986. LSE chief executive Xavier Rolet has insisted that the exchange, once a monopoly, will deliver record speed and stable trading in order to fight back against the fast erosion of its dominant marketshare by specialist electronic rivals."
Should it be...? (Score:3)
based on Novell SUSE Linux technology
Should it be Attachmate (err... Microsoft [computerworld.com]...?) Linux Technology already?
Re: (Score:3)
Even if Microsoft bought some rights from Novell via the Attachmate deal, the current stack is *not* Microsoft tech. The Accenture/MS solution (TradElec) was binned as it was problematic, it crashed for a whole day [practical-tech.com] and it never reached its performance targets (using Server 2003 and SQL 2000 [computerworld.com]). The CEO who had brought it in was apparently fired [computerworld.com].
London Linux (Score:5, Funny)
Re:London Linux (Score:5, Funny)
Does this mean the prompt will be a GBP (£) sign instead of a dollar ($) ?
No, but the root prompt will be £ instead of #.
Re: (Score:2)
Re: (Score:3)
Re: (Score:2)
Ha, talk about mental gears clicking into place. I'd grown so use to seeing the dollar sign as a prompt I'd completely separated it from the currency in my mind, as if there were two different symbols.
the problem is algorithmic trading... (Score:4, Interesting)
the exchange, once a monopoly, will deliver record speed and stable trading in order to fight back against the fast erosion of its dominant marketshare by specialist electronic rivals
The issue facing markets isn't that. It's algorithmic trading:
https://secure.wikimedia.org/wikipedia/en/wiki/Algorithmic_trading#Issues_and_developments
https://secure.wikimedia.org/wikipedia/en/wiki/2010_Flash_Crash
Re:the problem is algorithmic trading... (Score:5, Insightful)
What makes you think there needs to be exactly 1 problem?
one problem (Score:2, Insightful)
Re:one problem (Score:5, Insightful)
Blaming an economic collapses on greed is like blaming a bridge collapse on gravity. Both are always there, you design your system around them.
Re:one problem (Score:4, Insightful)
Blaming an economic collapses on greed is like blaming a bridge collapse on gravity. Both are always there, you design your system around them.
Yes, and part of your system is not to encourage the greediest fuckers so much, and certainly not to let them escape oversight until it all goes tits up.
Re: (Score:2)
So, you essentially confirm what he said, you just offer advise how to deal with it. Greedy people are the problem.
To go into your analogy: the purpose of the bridge is to allow the economy to grow and be more efficient, not to provide some very smart people with an implement that stores a huge amount of potential energy due to its intended position in the gravitational field of planet Earth. Protect the bridge.
The market is there to provide companies with capital, so that the economy can grow. Protect it.
Re:the problem is algorithmic trading... (Score:5, Funny)
The is only one problem: entropy, leading to the eventual heat death of the universe.
Re: (Score:2)
The is only one problem: entropy, leading to the eventual heat death of the universe.
Yes, that sort of puts everything in perspective. If you're immortal.
Re: (Score:3)
Dick Clark has been complaining about it since the original new year's eve.
Re: (Score:1)
Re: (Score:3)
You couldn't be more wrong. The absolutely opposite is true. Algorithmic trading is exactly what the exchanges want and here's why:
Increased trading means increased fees. Exchanges charge a fee for each and every order. MTFs such as Chi-X and Turquoise grabbed market share by having a clever pricing structure based on whether your order is sitting on the order book and filled (passive) or your order is immediately filled against something already
Impending Merger (Score:3, Informative)
This likely means that the Toronto Stock exchange will soon be using Linux as well [www.cbc.ca], if they aren't already.
Re:Impending Merger (Score:4, Informative)
The trading engine of the Paris stock exchange (previously Euronext, now NYSE-Euronext) is already on Red Hat Linux since 2006 [atoseuronext.com].
YES!!!! :) (Score:2)
I know it's not my most sophisticated comment, but I like open source so I am happy :)
Re: (Score:1)
Who said it will be open source? It can be based on Linux, and use proprietary software.
Re: (Score:1)
I can promise you what's installed there is far from open source.
Re: (Score:2)
I know it's not my most sophisticated comment, but I like open source so I am happy :)
The OS sitting at the bottom of the software stack is open source. The rest isn't. I imagine Linux is attractive because it supports a wider array of hardware than Windows would especially for a system which must guarantee low latency, high throughput.
Re: (Score:2)
Doubt it. It's not like you're going to cobble together the systems running a stock exchange from whatever you could find leftover in the warehouse.
Infinitely more likely that there's only a handful of software development houses with the expertise to develop stock exchange systems of this sort of size. (Which makes sense, as there's only a handful of big stock exchanges). Which means there's only a handful of companies that are going to be taken seriously if they bid for the contract.
Assuming there's on
Re:YES!!!! :) (Score:5, Insightful)
My understanding is the previous system was .NET over windows which raises large question marks over performance (e.g. unexpected garbage collections) and the sort of hardware that it could be run on.
Re: (Score:2)
My understanding is the previous system was .NET over windows which raises large question marks over performance (e.g. unexpected garbage collections)
It seems that the new system is running on Java, though, which is hardly an improvement in hard real-time department.
Re: (Score:2)
Well there are JVMs implemented in hardware and JVMs with real time RTSJ support (e.g. Websphere Realtime [ibm.com]. I suppose it's also possible that some parts of the system are Java and some aren't and I wouldn't really know what the LSE does without looking.
I know when I worked for a financial house that it was a mix of code. C++ / .NET represented the thick client code. Java represented the we
Re: (Score:2)
Yeah, and then another 3 years hiring somebody new and letting them rewrite it because your homebrew hobbyists who you thought would save you money low-balled the project because "LOL HOW HRD CUD IT B RLY?"
Nothing happens in the financial services industry in "about 6 months." Your project estimate reads like a PHB asserting, "I don't understand this, so I'll assume it's trivial, and can be done by half a dozen guys in half a dozen months."
Re: (Score:2)
Larry Ellison of Oracle once told Netscape's board that his cat could write a web browser. Microsoft popped out Internet Explorer later that year.
Re: (Score:2)
I'd make three comments in response:
1) MSIE 1 was started, by the accounts I can find, in mid-to-late 1994, and released in August 1995. So a year+ of development.
2) MSIE 1 was significantly simpler than a high-volume, high-availability critical trade processing system which handles hundreds of thousands of trades each day;
3) IE largely succeeded over time on the back of Microsoft's anticompetitive practices, and THEN would have laid in a corner neglected if Firefox hadn't come along to shock them out o
Re: (Score:2)
Actually, these types of things are usually in house. If there ever was a case for non-off-the-shelf software, a SE booking & matching engine is it. Most of the developers will be inhouse. The reason to use Linux is most likely that you can strip out all non-essentials. A path that an order takes from external interface to execution can be seriously streamlined with little intereferance from the OS. Imagine having fat kernels like Windows in the interfaces/perimeters and slim kernels almost akin to
ACN FTW (Score:3, Interesting)
And so ends one more of an increasingly long line of Accenture / MSFT snafus.
Re: (Score:3)
And so ends one more of an increasingly long line of Accenture / MSFT snafus.
At $40 million / snafu they couldn't care less. Now let's see the people at LSE who proposed the Accenture / MSFT solution. Will they be fired? Will the $40 million be discounted from their bonuses? I doubt it. Because they are probably working at another victim by now. Where will the next snafu be? Stay tuned.
Re:ACN FTW (Score:4, Informative)
Re: (Score:1)
Back in the old days when Pud ran FuckedCompany.com, he would love to post stories about Accenture. He'd always say: Accenture (pronounced ass-enter).
Re: (Score:1)
Is Novell SUSE Linux good?
I liked old SUSE, it was one of the first distro flavors
that I preferred over the stodgy (at that time) other
players. Haven't tried newest.
My old job uses OpenSUSE which I was pretty much
impressed with. (not the distro, the fact that my company
used it). What was presented was a locked down environ
that I didn't get to see too much of, til things went bad.
At prompt... everything was pretty much like you'd expect.
So, did you just want to know if the *dm is pretty? =)
cause the kernel and command set are relati
Re: (Score:2)
IMO the big things with a linux distro are
1: what is their release frequency and security update policy like? fast releases with little overlap mean a lot of time spent upgrading systems. Slow releases mean you run into problems with application software depending on stuff that is newer than your distro supplies.
2: what is their QA like? too long and some people complain it's old before it's even released. Too short and you get a buggy release that breaks in a lot of configurations.
3: what is their package
Re: (Score:2)
Novell SUSE Linux is pretty good as long as you think before you do stuff. Its easier to get into dependency hell than with RedHat or Debian/Ubuntu but other than that its very stable and nice. Personally i prefer Debian but i do manage a couple of SUSE machines taking serious loads day and night on a slew of different serivces and they have been working flawlessly without any problems at all.
Even Higher Speed! (Score:1, Troll)
Because clearly, ever more HFT is exactly what everyone needs!
Re:Even Higher Speed! (Score:4, Informative)
If you are an automated trader then you do need any speed you can get. If you aren't - well, you probably don't have any business being on the stock exchange any more. A bit more or less speed is not going to make any difference; you're hopelessly outgunned either way.
Re: (Score:2, Interesting)
Maybe I'm not cynical enough, but all the low-latency trading and algorithmic competition in the world isn't really going to change the fact that I can sit here with delayed stock quotes, look at a company's financial statements, look at the company's P/E ratio and potential for growth and put a bid in at a price of my choosing.
Re:Even Higher Speed! (Score:5, Interesting)
It sucks money out of the stock market.
Fans of it will say it provides 'increased liquidity'.
Me, I say, it sucks monkey out of the stock market. If it didn't the HFT people wouldn't bother doing it. The money comes from somewhere, and that somewhere is other investors. If it *doesn't* come from somewhere then creating it means there's more money and it comes from everyone via inflation.
That's my take. May be wrong, may be dense, but that's my take. Me, I'd scale back the whole thing massively because I still haven't had anyone explain adequately to me how, after they've gone public, the company's stock market valuation matters (to the company) for anything at all, except for perhaps their ability to rack up debt.
Re:Even Higher Speed! (Score:4, Informative)
"increased liquidity" has a cost.
If you want to buy now, or sell now, instead of spending a few days shopping for a willing partner on the other side of the trade you won't get as good of a price. But you also have the reduced risk of not spending a few days trying to unload your stock.
The one thing that you will notice is that with the high frequency traders the spread is now very small, but the volitility is much higher, so in some ways this is a reaction to discount brokerages and the decimalization of stock prices, as the fees are not in the spread but in the volatility. I don't know which is more profitable to the house, but the reduced spread and the greatly reduced commissions had to be made up for someway. Banks are more profitable now, but there are many fewer banks, so direct comparisons with thirty years ago are not possible without correcting for industry consolidation.
An example of the cost of liquidity is that, while exchanges are liquid, someone selling five to ten percent of a company in the open market in a few minutes can easily erase more than half of the companies market capitalization. But the stock is sold and the books are closed.
One difference between specialists and high frequency traders is that specialists have to be the buyer and seller of last resort, while HFT's have no obligation to make a losing trade.
I have not looked at the math behind technical trading recently, but in the nineties there were several papers published that suggested that technical trends could only see about 15 minutes into the future, and the predictive value more than five minutes into the future was murky, but not useless.
As far as your question of how a companies market capitalization matters to the company, it depends on the company. Some companies carry a large number of authorized shares on the balance sheet, but do have not issued them. The company can give bonuses as stock options which if the options are deep in the money, the investors are essentially paying a substantial percentage of employee compensation, instead of the company paying it. A high market capitalization does not guarantee that people will loan you money, but it does tend to lower your interest rate, and allow you to issue really long term debt. (Disney and Coke issued 100 year bonds, Canadian Pacific issued 1000 year bonds.) One other thing that companies can do with an over inflated market cap is buy things with the stock, see aol buying time warner as a famous example of that. I am sure that there are uses that I am forgetting at the moment. But I hope that is somewhat illuminating.
Re: (Score:2)
If you're a market maker - who may have a legal & contractual obligation to keep offering trades - yes, you do stay in the market until the trading curbs halt a complete meltdown if things are headed south.
Re: (Score:3)
I still haven't had anyone explain adequately to me how, after they've gone public, the company's stock market valuation matters (to the company) for anything at all, except for perhaps their ability to rack up debt.
If the company's stock is highly valued they can finance growth by selling shares. Otherwise they would have to "rack up debt" in order to grow.
Re: (Score:1)
Me, I say, it sucks monkey out of the stock market. If it didn't the HFT people wouldn't bother doing it. The money comes from somewhere, and that somewhere is other investors. If it *doesn't* come from somewhere then creating it means there's more money and it comes from everyone via inflation.
The whole point of the stock market is it's not a zero-sum game - the companies you invest in are creating wealth (by building a better mousetrap, from materials and labour that are worth less than the finished product), which goes to their owners, i.e. the stockholders.
Now, you're right in that the money that HFT guys make comes in the immediate from other investors. So flip it around: if all they're doing is creaming off profit, why would anyone ever buy from them or sell to them. Answer: they give you th
Re: (Score:1)
Stock trading is a zero sum game.
No it isn't; the total value in the system changes over time - the dow jones or whatever tends, in the long term, to rise.
For every buyer there is a seller, for every winner, there must be a loser.
No; mutually profitable trades can be made when something is worth more to the buyer than it is to the seller. That's the fundamental basis of capitalism.
Re: (Score:2)
You can. And as long as you treat the stock market as a longer-term investment - a place to park your money - the HFT activity isn't going to matter to you. Buy a portfolio or shares in a couple of index funds, then revisit your holdings oh, once a year or so. No problem. In a ten or twenty year time-frame you'll have done much better than the mattress or a bank account, and likely be ahead of bondholders.
But trying to play the stock market - to trade your way to wealth - is a losers game today if you're an
Re: (Score:2)
If it's in an index-linked fund there's a good chance you'll wind up beating automated traders. 80% of the things that try to beat the market average fail, sometimes spectacularly.
Re: (Score:2)
LOL. over the last decade or so, funds have (in general) been a great way to make next to no money at all. The dot-com burst and then the "crisis" we're apparently still recovering from.
Re: (Score:2)
Looks like somebody doesn't know about dollar cost averaging and long term investment in a diversified portfolio.
Re: (Score:2)
Actually, no, but then the dollar is meaningless to me as I don't live in the US.
And you can diversify as much as you like, roughly every ten years (it seems) everything takes a major dive for a year or two.
Re: (Score:2)
Pound cost averaging? Unit cost averaging? Cost average effect? Any of those ring a bell? They all (along with Dollar cost averaging) describe the same thing.
And "things taking a dive" is exactly *why* you diversify - across sectors, across regions, and across instruments, and across company sizes. In laymen's terms, this avoids "putting all your eggs in one basket." Some sectors of the economy hold up better in downturns than others; Some instruments hold up better than others; some regions hold up
Re: (Score:2)
Looks like somebody doesn't know about dollar cost averaging and long term investment in a diversified portfolio.
Fuck me, Oscar Wilde is alive and well and posting on slashdot.
Re: (Score:2)
I'll assume you're aware that dollar cost averaging and diversifying a portfolio for long term gains are accepted (and proven) methods for investing that greatly eliminate the volatility of your portfolio in what is, essentially, a cyclical market. Investments over the last decade have absolutely made money if you spent a few days learning how to invest, and then a few hours each quarter balancing your portfolio to match your plan.
The stock market is not a "this week, this month, this quarter" endeavor. I
Re: (Score:2)
One social benefit of HFT is putting day traders back to real work.
get the fact (Score:5, Informative)
Re:get the fact (Score:5, Informative)
Yes, it was one of them. I worked on another Reuters Intelligent Advisor which ran like a 3 legged dog, a very expensive dog, until someone did the decent thing and shot it through the head.
I don't think RIA's expensive failure can be wholly blamed on .net. I think the technical team deceived management and probably themselves about what they could do. They had drunk the SOA/Web Services kool aid and the architecture was basically wrong. I suspect a number of devs saw the project as resume keyword fodder.
Re:get the fact (Score:4, Interesting)
Still, Microsoft was heavily involved on every levels of the project. If Microsoft cant make .net / MSSQL work and design it right, who can?
Re: (Score:2)
Jimmy Hoffa.
Re: (Score:2)
Who from Microsoft? Are you talking about the intern that graduated college 6 months ago and got assigned the project, or are we talking about someone who actually knows what they are doing? You talk as if suddenly Microsoft turned the entire company and made making that project work their #1 priority, which isn't nearly the case.
I can have a team write a similar version on linux and (pick your language) that does the same thing, only worse with more bugs and slower, as well. It's really not that hard.
Re: (Score:2)
I think its safe to assume that MS are NOT going to put a trainee on a project that involves one of the worlds major stock exchanges that would be a major showcase for their technology.
Re: (Score:1)
There is one person, but God does not code in C#. She prefers Assembler.
Re: (Score:2)
Highly Reliable Times (Score:1)
I know of a newspaper that won't be publishing this story...
Stable by insistence? (Score:2)
I'm not sure, even if a senior manager insists on stability, that it is guaranteed to happen. What he could do is to insist that, if any instability occurs in the new system, the workforce positions will also become unstable.
"Runs on Linux" But what is the app written in? (Score:1)
I hear a lot of fanfare about the OS, but what is the danged trading app written in? You can't "write something in linux"... is it C, Java, COBOL what?
Re: (Score:2)
running on SUSE Linux (Score:1)
> At 8am today, the exchange’s main venue went into live trading with the Millennium IT matching engine, developed in C++ programming language and running on SUSE Linux ..
Why did they choose SuSE and how does Microsofts intellectual property patent covenant with Novell impact on the decision. Is part of what the LSE paying to Novell being funneled back to Redmond. Is this the future of Microsoft innovation ?
LSE investigating potential IT problem at close (Score:3, Interesting)
Hope not.. (Score:2)
Re: (Score:1)
Re: (Score:2)