London Stock Exchange Woes not Linux's Fault

London Stock Exchange Woes not Linux's Fault

Summary: The London Stock Exchange is having problems both its systems and on its Web sites, but it's not Linux's fault.

SHARE:

The London Stock Exchange (LSE) has had better times. First, it had reoccurring problems with its integration with large-stock market data players such as Thomson Reuters Eikon, Interactive Data, and Morningstar. Then, adding insult to injury, Google rightfully flagged the Exchange's site as a malware danger, thanks to a third-party advertiser that was hosting malicious software. None of this had anything to do with the LSE switching to Linux as the basis of its new trading system.

I mention that because I've gotten several obnoxious e-mail messages claiming that all of LSE's troubles are Linux's fault. Sorry guys, it's not. Here's what, as far as can be told, is what's really going on.

First the malware problem actually belongs to a third-party ad vendor. If you clicked on this company's ads on the LSE Web page, or any other Web page, you'd end up at a site bearing Windows malware.

Google, which does a good job of finding and marking dodgy sites, marked the LSE site as having malware problems over the weekend. Users of Web browsers, such as Chrome, Firefox, and Safari that use Google's Safe Browsing API (Application Programming Interface) were warned off the LSE site. Users of Internet Explorer, which doesn't use this API, could have still visited the site without any warning.

Since the problem was uncovered, the LSE has removed the advertiser and its infected links. Google now reports the site as being safe for browsing.

This kind of thing can happen to any Website that uses third-party ad services. Still, I do have to ask what the LSE was doing letting any ads on its site in the first place. I mean, it's a stock exchange! They don't need to run ads to make money.

The other problem is far more serious. Beginning right after the switchover to the new system, the LSE started having troubles sending data to three of the big four traders. Bloomberg and most of the smaller trading partners seem to have done their homework and are having no trouble with the new LSE system.

On Friday, the companies that didn't have their ducks in a row and the LSE combined to bring the Exchange down for four hours. Tomorrow, March 1st, the big four, some smaller trading houses, and the LSE will be getting together to work out the data transfer glitches.

Specifically, the problem seems to be in differences in the ways how each of the major stock dealers, Thomson Reuters Eikon, Interactive Data, and Morningstar, handle the Financial Information Exchange (FIX) protocol and its closely related FIX Adapted for Streaming (FAST) protocol. It's noteworthy that Bloomberg, which doesn't use FAST, has not has the problems reported by the others which do use FAST.

The FIX Protocol Ltd. has stated that they've not been told that the problem lies with FIX. FIX, though not well known outside the trade technology field, has long been the stock market trade execution and data protocol. FAST, which is used to support high-throughput, low latency stock data communications and trades, has also been widely adopted, but not nearly as widely as plain old FIX. So, what happened?

At an educated guess, it looks to me like the LSE implementation of FAST isn't compatible with Thomson Reuters Eikon, Interactive Data, and Morningstar versions of FAST. This kind of data transfer protocol interoperability bug shows up all the time.

What usually doesn't happen though is that a bug of this magnitude involving such large players as the LSE and these financial firms would have stayed hidden during the 18-months of testing before Millennium Exchange was rolled out. While I wonder exactly how the LSE implemented FAST and its associated networking technologies, I also wonder how in the world the trading companies couldn't find the incompatibilities earlier. It looks to me like they'll plenty of blame to spread around in tomorrow's meaning, but none of it will fall on Linux.

Topics: Software, Banking, Linux, Malware, Open Source, Operating Systems, Security

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

160 comments
Log in or register to join the discussion
  • I smell sabotage

    from sleeper M$ agents.
    Linux Geek
    • I smell a coverup from SJVN.

      @Linux Geek "It's never Linux's fault, it's always something else - Linux never has problems"

      I think he's Dietrich T. Schmitz's alter-ego. He appologizes for Linux just as much!!
      Will Farrell
      • SJVN spins faster than a tornado

        Gee, that's some hard core FOSS lip stick registered 10 out of 10 on Donnie's scale.
        LBiege
    • I smell apologies and apologists

      @Linux Geek
      Never attribute to malice whan can adequately be explained by incompetence.

      In this case it is the incompetence of the Linux zealots on the LSE. Despite dealing themselves all the cards by testing for a full year, <b>Linux and the zealots still managed to screw it up</b>.

      And now the apologists like SJVN come crawling out of the woodwork. The very same people who were all over LSE when they experienced network problems. That HAD to be Windows' fault!
      honeymonster
      • RE: London Stock Exchange Woes not Linux's Fault

        @honeymonster Such a troll you are. The problems have nothing to do with the OS. These problems are typical of migrations and compounded by application level problems.
        nicholas22
      • See my post below.

        @nicholas22: [i] The problems have nothing to do with the OS.[/i]

        The one titled "Nor were their previous problems Windows fault."

        What's good for the goose...
        ye
      • RE: London Stock Exchange Woes not Linux's Fault

        @honeymonster That's because those really were Windows problems. This isn't an operating system problem. Simple really.

        Steven
        sjvn@...
      • I just love the way he spins it

        @honeymonster

        It's the fault of the people who didn't change anything. They should have done whatever to match whatever the LSE was changing. Or perhaps they should have just matched the data exchange that was already happening, which would have been the smart thing to do.

        As Linux was part of the vendor supplied solution, it's part of the problem. It's likely the software built on top of it but it's sold as an entire solution.
        LiquidLearner
      • You have proven neither.

        @sjvn: <i>That's because those really were Windows problems.</i><br>
        So you keep saying but have yet to prove.<br><br><i>This isn't an operating system problem.</i><br><br>So you say but have yet to prove.
        ye
      • RE: London Stock Exchange Woes not Linux's Fault

        Steven,
        Enjoyed the article.

        Hooah!
        choyongpil
      • Dear Steven, Linux apologist

        sjvn@

        1) You have the old system which went down for 8 hours. The LSE never fully explained what actually happened, but they *did* say that it was a <b>network connectivity problem</b> and that it was <b>unrelated to the TradElect</b> platform.

        <b>Yet you proceed to conclude that it was the fault of Windows and .NET (on which TradElect was built).</b>

        2) We now have a sparkling new system, which has been designed to run on Linux, has been tested for 14 months but when stressed by real-world conditions <b>starts to convey faulty prices</b>. Granted those faulty quotes are delivered *really* fast. By all accounts this is *not* just a problem with external suppliers as even LSEs own quotes on their own trading stations are wrong. From insiders (read slashdot) we know that the app is rife with race conditions.

        <b>Yet you proceed to absolve Linux entirely for this fiasco</b>

        The reality is that Linux has an abysmal record on multi-threading. Many libraries - and yes also kernel ones - are still <b>unknown territory</b> when it comes to true multi-threading and asynchronous operations - what is exactly what we are talking about here.

        The decision to go with Linux was accompanied by a decision to code in C++ as well. Modern processor architectures make it extremely difficult to get multi-threaded or even multi-process and especially asynchronous programs correct. The problem is compounded by processor caches etc, where memory may not be written through and thus other cores/processors may see something different at the very same memory locations.

        Higher level execution environments like .NET and Java trades a little performance for a lot more consistency by introducing e.g. "barriers" - whenever a synchronization primitive is used it is also a memory barrier. When going C++ and raw Linux libraries you are without this guidance. You are on your own. An obviously the Linux zealots at LSE managed to mess it up.
        honeymonster
      • RE: London Stock Exchange Woes not Linux's Fault

        @honeymonster Faulty prices can't be GNU/Linux fault in any way. Also, fact that it have been tested for long tome with no problems shows that there is nothing wrong with system. It implies sabotage.
        gnufreex
      • RE: London Stock Exchange Woes not Linux's Fault

        @honeymonster


        "The reality is that Linux has an abysmal record on multi-threading. Many libraries - and yes also kernel ones - are still unknown territory when it comes to true multi-threading and asynchronous operations - what is exactly what we are talking about here."

        This is just crazy talk. NYSE has no problems, same goes with Qatar stock exchange and many others. NYSE has bigger systems than LSE. So crackpot stroy about "unreliability" of GNU/Linux doesn't hold any water.

        Interesting fact is that only LSE uses SUSE and everyone else uses RHEL. SUSE is from Novell. Novell is company that is in bed with MSFT and which took MSFT money to promote OOXML faux-standard (see http://www.groklaw.net/article.php?story=20101219121621828 ) which is completely against interests of just about anyone other than Microsoft. Including Novell investors. Sabotaging their own platform, for some amount of $$$ paid to top brass, is something to be expected from Novell.
        gnufreex
      • RE: London Stock Exchange Woes not Linux's Fault

        @honeymonster <br>From insiders? Please, enlighten us more... or perhaps, could it be... you are vaporware?<br>"The decision to go with Linux was accompanied by a decision to code in C++ as well. Modern processor architectures make it extremely difficult to get multi-threaded or even multi-process and especially asynchronous programs correct. The problem is compounded by processor caches etc, where memory may not be written through and thus other cores/processors may see something different at the very same memory locations."<br>Really??? I am running multithreaded jobs (32 proc per job) and have never seen that.<br>But perhaps you are right and this is the reason all supercomuters (actually close to 90%) of those run <sarcasm>Windows</sarcasm> ;-).
        kirovs@...
      • RE: London Stock Exchange Woes not Linux's Fault

        @gnufreex

        'NYSE has no problems'

        http://www.nyse.com/pdfs/Arca_Outage_10_29_07.pdf
        http://www.nyse.com/pdfs/Arca_Outage_06_12_08.pdf
        http://www.nyse.com/pdfs/Outage_Update_20090303.pdf
        http://www.nyse.com/pdfs/Outage_04172009.pdf
        http://www.nyse.com/pdfs/Outage_08052009.pdf
        http://www.nyse.com/pdfs/Update%20on%20Routing%20Destination%20Tag%2076%20arca.pdf
        http://www.nyse.com/pdfs/Client%20Notice%20outage%20amex%2005_06_10.pdf
        http://www.nyse.com/pdfs/Quote_Interruption_6.10.2010.pdf
        http://www.nyse.com/pdfs/Trader%20Update%20Amex%20outage%20MNOQ.pdf
        http://www.nyse.com/pdfs/Trader%20Update%20amex%20outage.pdf

        With respect, when you haven't a clue about the topic, it's best to keep quiet. This applies equally to Vaughan-Nichols, I'm afraid.
        WilErz
      • RE: London Stock Exchange Woes not Linux's Fault

        WilErz,
        Woo Hoo! All are terrified now.

        Steven,
        Great article.
        choyongpil
      • @ye, neither do you

        [i]So you keep saying but have yet to prove.[/i]

        The burden's on you to prove otherwise.

        [i]So you keep saying but have yet to prove.[/i]

        The burden's on you to prove otherwise.
        LTV10
      • Burden of proof?

        @LTV10

        Since when has the burden of proof shifted to anyone other than the person making an argument? Especially someone making that argument in a supposed "journalistic" role?
        daftkey
      • RE: London Stock Exchange Woes not Linux's Fault

        @WilErz
        I think most of these are hardware failures, how is this relevant? Not to mention that the longest PARTIAL outage was 34 min. Pretty impressive I think. Did you actually read the reports?
        kirovs@...
      • Yes, burden of proof

        Since @ye didn't counter @sjvn@...'s argument with an argument of his own.

        Or haven't you been tracking this discussion...
        LTV10