Forgot your password?
typodupeerror
Networking Australia Communications Hardware IT

Australian Company Promises Switching Hardware With Sub-130ns Latency 77

Posted by timothy
from the time-to-start-straightening-the-undersea-bits dept.
snowdon writes "The race for low-latency in finance and HPC has taken a major turn. A bunch of engineers from Australia have 'thrown away the air conditioning' in a traditional switch, to get a 10G fibre-to-fibre latency of less than 130ns! Way faster than more traditional offerings. This lady (video) would tell you that it's equivalent to just 26m of optical fibre. Does that mean we just lose money faster?"
This discussion has been archived. No new comments can be posted.

Australian Company Promises Switching Hardware With Sub-130ns Latency

Comments Filter:
  • by Anonymous Coward on Tuesday June 05, 2012 @10:48AM (#40219415)

    The trick was not reading TFA.

  • by CanHasDIY (1672858) on Tuesday June 05, 2012 @10:52AM (#40219463) Homepage Journal
    Now they can suck our economy dry with even more speed and efficiency!
    • by Anonymous Coward

      Now they can suck our economy dry with even more speed and efficiency!

      I'm sure they'll have some nice propaganda to tell us all about how this benefits the common man.

      Enriching ourselves without producing anything of value is good for you, really! Now grab your ankles and prostrate yourselves before the financial machine again...there there... that's a good subject I mean citizen. Man financial slavery is much easier than the old way, why intimidate when you can brainwash? Now bust your ass at your low wage job to create more value for us, and be thankful you even have

      • Now they can suck our economy dry with even more speed and efficiency!

        I'm sure they'll have some nice propaganda to tell us all about how this benefits the common man.

        They already do; it's called the 2012 Presidential Election. [opensecrets.org]

        Donations to Barack Obama in 2008 - $1,013,091
        2012 (reported thus far) - $80,224

        Donations to Mitt Romney (2012, reported thus far) - $528,180


        I believe this is what they in the finance 'industry' refer to as "hedging."

  • by Anonymous Coward

    Pfft, I can't win Quake matches with that latency.

  • Darn work and filtering. ;-) Oh well..... I'll bookmark and watch it when I get home 11 hours from now.

  • Lots of HPC installations use InfiniBand. (Go here [top500.org] and select "Interconnect" or Interconnect Family" from the dropdown menu to see the stats.) According to its wiki entry [wikipedia.org],

    The single data rate switch chips have a latency of 200 nanoseconds, DDR switch chips have a latency of 140 nanoseconds and QDR switch chips have a latency of 100 nanoseconds.

    Can these numbers be directly compared to the "fiber-to-fiber latency" reported in the article summary?

  • by the eric conspiracy (20178) on Tuesday June 05, 2012 @11:15AM (#40219789)

    This is not good news; it promotes a trend in a technological approach to making money in the stock market that should be flat out illegal.

    • by vlm (69642) on Tuesday June 05, 2012 @11:38AM (#40220047)

      This is not good news; it promotes a trend in a technological approach to making money in the stock market that should be flat out illegal.

      Why? I know what it is, how it works, how it makes money, and I'm not overly concerned. So they make a market, and they make it really fast. Eh.

      All I ever hear as a rationalization for dislike of it is "workers of the world unite" and "1% OWS" and "something bad happened to the markets, I don't like witches, so lets hang the witch because she has property I like that we can take from her" and "I hate the rich" and all that sort of stuff. I never hear a good moral or ethical explanation of why doing what you normally do, but really fast, is so wrong.

      Another good question, is assuming you "make it illegal" how in the world would you enforce that? What would the reg look like? Like... your connection to the market must be this far away blah blah even if it means making a 1000 KM coil of fiber on the data center floor, or maybe all traders must connect to "the market" over geosync satellite links so the average latency is both long and more or less equal?

      Please don't confuse conventional HFT with "Flash trading" which is basically window dressed up insider trading, corrupt as all heck. If you do insider trading and trade your own book etc thats just illegal, but if you do it really fast people like to pretend its just "flash trading" and not wrong.

      The easiest way to "get rid of" HFT is to fix decimalization. Back when we traded on eights thats too wide to make money on a HFT strategy, and if we allowed millionth of a penny trading quotes that would narrow bid/ask resulting in too little money generated by HFT. Decimalization seems to be almost the perfect pricing system to result in massive HFT strategies, so if you really hate HFT (and WHY?) then just fix the pricing structure so its not profitable anymore. Rather than playing games with legal enforcement of favored or disfavored behavior, simply make disfavored behavior unprofitable. I suppose the opposite solution of going big would work just as well, fine we'll only trade stocks on dollar values now no fractions of a dollar.

      • by Nethead (1563)

        I never hear a good moral or ethical explanation of why doing what you normally do, but really fast, is so wrong.

        The nuns back in grade school gave us moral reasons not to wash "down there" really fast. Although through much "scientific discovery" I concluded they were wrong (or just trying to keep a good thing for themselves.)

      • by erbbysam (964606)

        The easiest way to "get rid of" HFT is to fix decimalization. Back when we traded on eights thats too wide to make money on a HFT strategy, and if we allowed millionth of a penny trading quotes that would narrow bid/ask resulting in too little money generated by HFT. Decimalization seems to be almost the perfect pricing system to result in massive HFT strategies, so if you really hate HFT (and WHY?) then just fix the pricing structure so its not profitable anymore. Rather than playing games with legal enforcement of favored or disfavored behavior, simply make disfavored behavior unprofitable. I suppose the opposite solution of going big would work just as well, fine we'll only trade stocks on dollar values now no fractions of a dollar.

        Don't forget that it was possible to game they system in different ways when they traded on fractions. Changing the decimal system seems a bit overkill(also, I'm still a bit confused how that wouldn't create further issues?) when a small tax on ultrafast trades/a small delay on the stock exchange end would solve the problem created by these HFT machines.

        • by vlm (69642)

          also, I'm still a bit confused how that wouldn't create further issues?

          The HFT idea is to "win" maybe 1% of time on a 50% of the bid ask spread times number of shares traded by retail non HFT investors.

          Assume the non-HFT quantity remains the same.

          Using penny level decimalization, the bid ask difference will tend toward... a penny. So 1% of 50% of a penny times a billion shares traded per day is $50K per day, optimistically (assuming you "win" as much as 1% of the time, probably much less). Times 200 stocks, that's $10M revenue per day well worth building the infrastructure t

          • by erbbysam (964606)
            Wouldn't a penny "transactions fee" or "tax" solve this issue instead of just pushing it further down the line? Also, what's to stop somebody from doing even more transactions to make that $1000 into a much larger number?
            • by vlm (69642)

              A penny per share tax is a buck on a round lot trade of 100 shares to kill HFTs. The problem is I don't lose a buck to HFTs but I would lose a buck to the govt. Its just a cheesy stealth taxation plan. Blame the HFTs, rich people suck, horray for the 99% we're gonna destroy HFTs, but the only real end result is the govt takes more of my money. So a transaction tax is a great way to cause a couple orders of magnitude more damage to the economy than HFTs could theoretically ever cause.

              This still doesn't e

              • by erbbysam (964606)

                A penny per share tax is a buck on a round lot trade of 100 shares to kill HFTs. The problem is I don't lose a buck to HFTs but I would lose a buck to the govt. Its just a cheesy stealth taxation plan. Blame the HFTs, rich people suck, horray for the 99% we're gonna destroy HFTs, but the only real end result is the govt takes more of my money. So a transaction tax is a great way to cause a couple orders of magnitude more damage to the economy than HFTs could theoretically ever cause.

                Who's to say this money would go to the government? Don't exchanges need to pay huge amounts of money for infrastructure to support this activity? Also, isn't there a clear segregation between "normal" traffic and this high speed traffic?

                This still doesn't explain the fixation on killing HFTs. So they get a "sin" tax. On what sin, exactly? nobody knows.

                Because financial engineering has just been so beneficial to society... /s http://en.wikipedia.org/wiki/High-frequency_trading#Effects [wikipedia.org] http://en.wikipedia.org/wiki/2010_Flash_Crash [wikipedia.org]

                WRT increasing transaction volume to make more than $1000... Remember that superman movie where Richard Pryor took all the fractional cents out of everyone paycheck, put them in a pile, and kept that substantial amount? Well HFT is like that in trades where a bunch of competing HFTs fight each other to, on long term average, determine the sub-cent price of a share priced in cents, and then, essentially, the HFT who skims off the least amount on top of the sub cent slice beats the greedier HFTs and keeps the extra skim. Well, in the movie, if Richard Pryor wanted more money, he'd have to make more transactions, make more paychecks. But Mr Pryor doesn't get to decide how many times paychecks are written per month (Weekly checks? Daily checks?), or how many people are employed... This tired old hollywood plot was recycled into Office Space, and probably many more movies. There's only like 50 movies, endlessly redone.

                Ya ya ya, we all know the old skimming scheme that's become possible because of electronics controlling money. Just because there is a way to manipulate a system, doesn't make it a good thing. 21 is a good (blackjack)movie, but will still get you thrown out of a casino. There's nothing to say that these HFT's aren't skimming money from regular investors and vice-versa. That said though, why does the market exist in the first place? It's not for a bunch of machines to race to the penny.

                Think about making a competitive system to determine the sub cent fractional price of shares traded at cent rounded prices. Read the wikipedia article below. You pretty much end up reinventing HFT.

                http://en.wikipedia.org/wiki/Simulated_annealing [wikipedia.org]

                Instead of sending me to a generic NP-Hard problem, I find this - http://en.wikipedia.org/wik [wikipedia.org]

      • There are several reasons as to why HFT should be illegal. And it doesn't have anything to do with your stupid strawman arguments regards OWS etc, or front running which has been illegal for some time.

        Some examples of the abuse that occurs are the practices of placing and rapidly cancelling limit orders to stimulate market orders on the other side, which are then executed at a less favorable price also placed by the HFT. In other words, HFTs are gaming the system to exploit sub-optimal behavior by slower tr

      • Another good question, is assuming you "make it illegal" how in the world would you enforce that? What would the reg look like?

        That's really easy: Nobody (and I mean nobody) gets access to flash orders. No stock exchange should be allowed to let a third party inspect the details of a pending trade, and then make their own transaction before the initial trade is complete.

        That is so painfully obvious I can't believe you would have to ask.

    • This is not good news; it promotes a trend in a technological approach to making money in the stock market that should be flat out illegal.

      I always thought that if those losers in Anonymous (or anyone else sick of Wall St. and their economic stranglehold) wanted to actually do something meaningful, they would create and release an open source HFT program anyone can use, and help the populace beat the banksters at their own game...


      nudge nudge...

      • The bankster's game has nothing to do with HFT. Regardless, you aren't going to beat HFTs with software, you also need physical proximity to the exchange's gateways. And that isn't cheap.
        • The bankster's game has nothing to do with HFT. Regardless, you aren't going to beat HFTs with software, you also need physical proximity to the exchange's gateways. And that isn't cheap.

          Well, if that's the case then I agree with OP's assessment that the process should be criminalized.

          • Why do you say that? How is hft different than any other business? I can't afford to build my own car from raw materials, but Pontiac can. I don't think you would suggest that car making should be illegal. You are free to open an HFT business if you can get the money together, just as you are free to start making cars if you can put the cash together.
      • by mikael (484)

        The banks choose cto locate their trading computer systems right next to (if not in) the co-location centers where the fibre-optic routers of the trading network are connected. The boxes being discussed connect those computers to the trading network. Everyone else gets the slow as molasses "Internet".

        The trading algorithms work by making frequent future price bids and sales, then cancelling those if the price isn't right. Then they can pocket the difference. Because of their proximity and short "hop" distan

        • by Bengie (1121981)
          Don't even care about seconds. There is a noticeable difference between being a block away and in the same building. When you're talking about sub-microsecond difference a noticeable difference, then it's just getting insane.
      • by gl4ss (559668)

        there's plenty of robo trading programs available for you if you wish..

        the problem is that you'd need to get preferential fast access to the market too for them to be any good for you.

        now - add the fact that it's the same entities providing the chance to trade these stocks which has that preferential treatment(bought with money, sure) and it does get a little fishy. they're running a business using information which at the point that they receive it is in fact insider information(obtained before others have

      • by makomk (752139)

        Even if they did create a HFT program that everyone could use, and even if it didn't require colocation in the exchange's server room thanks to the speed of light (which it does), and even if it did manage to outsmart the existing trader's software, the big traders would just find some way to have them arrested and convicted of fraud just like they did with the last group who tried that without inside contacts and political influence. You see, apparently it's fraud to do HFT in a way that confuses someone e

    • It would have helped in the Facebook IPO. All the investors with this level of link would have devalued Facebook stock faster than normal folks on DSL could have bought it.

      http://www.google.com/finance?client=ob&q=NASDAQ:FB

      Expand out the whole chart. Looks like a great design for a water slide at a kiddie park.

    • by jmerlin (1010641)
      I say we postpone any decisions like this until after they figure out how to send information faster than the speed of light. Greed is a pretty powerful tool, it appears.
  • Where you could previously only bailout one ot two banks per year, thanks to this technology about a dozen bailouts are now possible.... Isn't it nice.

  • Is like playing the Russian roulette game. With plutonium bullets. Close to a fuel tank.
    Yes, we'll all loose money much faster!

  • Meanwhile... (Score:5, Interesting)

    by bertok (226922) on Tuesday June 05, 2012 @12:02PM (#40220421)

    Even if a lot of research has been put into reducing the latencies of switching technology, the vast majority of real-world deployments are nowhere near where they should be. The result is corporations spending millions upgrading their core switching, and then the result is the same or worse than what they had with ordinary gigabit technology.

    I've been to more than half a dozen sites recently with new installations of 10 GbE, but with terrible network performance. All too often, I'm seeing latencies as high as 1.3 milliseconds even for servers on the same switch. Read up on bandwidth-delay product [wikipedia.org] too get an idea of why this would severely nerf throughput. The odd thing is that at a number of these sites, older servers with 1 GbE connected to the same switching infrastructure get 100% of wire speed without issues.

    I don't know exactly what the root cause is, but I'm starting to suspect that the extra latency is coming from somewhere that network engineers don't usually test, like CPU power management taking an excessively long time to wake up a core to respond to a packet. What I think happens is something like this:
    - The fast network delivers a burst of data very quickly.
    - The receiver CPU starts to slowly wake up from sleep mode, while the sender is waiting for an ACK packet, because it has finished sending an entire TCP window.
    - The sender CPU goes to sleep, because it still has nothing to do.
    - The receiver CPU finally gets around to the packet, which it processes quickly and efficiently, sending the ACK back.
    - The receiver OS sees that the CPU is "0.1%" busy, has nothing to do now, so it sends it to back to sleep.
    - The sender CPU starts to slowly wake up, while the receiver is also asleep, waiting patiently for more data.
    - The cycle repeats, with more waiting at every step.

    With slightly slower networks or CPUs, the CPUs never get a chance to be idle long enough to enter a sleep state, so everything is always ready for more data. I've seen 3x improvements in iPerf TCP throughput by simply running a busy-loop in a background process!

    • Re:Meanwhile... (Score:4, Interesting)

      by Cassini2 (956052) on Tuesday June 05, 2012 @12:35PM (#40220961)

      The number of interrupts per second that a modern processor can handle has stayed relatively fixed for a large number of years, and network response time is a strong function of interrupt performance.

      It has been a while since I benchmarked interrupt response on a variety of processors with the same block of code. However, when I last did it, a 32 MHz 8-bit PIC18 series microcontroller (MCU) was capable of processing a real-time interrupt at 10,000 times per second. A much faster 3 GHz x86 CPU could manage 70,000 interrupts per second under Real-Time Linux. This wasn't much of an improvement, considering the x86 CPU is theoretically 100 times faster, was capable of executing multiple instructions per clock, and the little 8-bit PIC MCU has a hard limit of around 8 MIPs.

      Why does this happen?
      For a variety of factors including:
      1. The amount of data a modern CPU must flush on an interrupt is huge. Most of the cache needs to be flushed and replaced, and this means a great deal of memory bus bandwidth.
      2. Main memory latency in the modern PC is huge. Modern compiler designers assume it to be on the order of 100's of clocks.
      3. Modern operating systems have to reload the page tables for every user/kernel mode transition. This takes time, and possibly more interrupts.
      4. The I/O response time of the PCI bus is terrible relative to the cycle time of the processor. This means that most of the network traffic is transferred via DMA. However, this optimization only helps in terms of bandwidth, not latency.
      5. Multiple cores don't really help. Synchronization overhead is a killer. Tricks like software transactional memory can help, however typical I/O devices lack the ability to support software transactional memory. Even simple techniques, like giving each core its own I/O and thus eliminating synchronization overhead, are often not implemented. Linux is still trying to eliminate the "Big Kernel Lock".

      In short, the interrupt response latency in modern processors has really not improved much. This was a big discussion in some of the Ethernet working groups, because some of the simulations pointed out that 10Gb Ethernet performance might be limited by interrupt response time on the processors.

      • by Bengie (1121981)
        There is hope around the corner, Netmap claims to share no locks among established TCP connections. It also claims to scale extremely well with many cores. A 950mhz single core CPU can saturate a 10Gb interface. It can also attach a given connection a to a given core and have the NIC forward the data to the L2 cache so it's ready to be processed. It is also capable of handling send/receive events on different cores for the same connection.

        Should make for some fast routers. It is being ported to Linux and
      • by joib (70841)
        Some additional points:

        - FWIW, Linux finally got rid of the BKL in the 3.0 release or thereabouts.

        - Many (most?) 10Gb NIC's are multiqueue, meaning that the interrupt and packet processing load can be spread over multiple cores.

        - Linux and presumably other OS'es have mechanisms to switch to polling mode when processing large numbers of incoming network packets.

        That being said, your basic points about interrupt latency being an issue still stand, of course.
      • Re:Meanwhile... (Score:4, Interesting)

        by bertok (226922) on Tuesday June 05, 2012 @06:19PM (#40226143)

        That's similar to what I'm seeing.

        I found a "DPC ping" tool which queues a simple Kernel task from User mode, and measures the response time as if it was an actual network ping.

        I found it interesting that it was well correlated with both ping times and tested network throughput, and that the DPC ping time wasn't consistent. Some fast CPUs had terrible DPC pings, and some older slower models were much faster. The operating system also contributes significantly, I found some combinations where the average was OK, but there were regular spikes of high latency.

        Now that I've done more benchmarks across more sites, I've been recommending jumbo frames much more often. Previously I though it was a nice-to-have, but these days I'm starting to be of the opinion that without it the money spent on "10 GbE to the server" is just wasted.

      • by gramty (1344605)
        As mentioned interrupt processing / context switches can add latency and jitter. As a 10G LAN has a very small round-trip delay (RTD), TCP can calculate the re-transmit timeout (RTO) to be a very small number. which makes it very sensitive to small amounts jitter, and can result in packets incorrectly being flagged as lost, this causes re-transmission and for the TCP window size to be zeroed (killing throughput) (some congestion control tuning can help here). The problem can become severe with 10G networks
  • Unfortunately, their hardware has the unfortunate side-effect of flipping all bits upside-down.

  • Well they already have loops of cables behind servers to ensure that no one could have a possible advantage. http://www.abc.net.au/news/2012-04-11/kohler-high-frequency-trade-parasites-at-heart-of-asx/3943052 [abc.net.au]

    Anyone with a mind set in reality will understand how didiculous that is!

Do not simplify the design of a program if a way can be found to make it complex and wonderful.

Working...