Genesis Releases Latest Ping Times Tests

Discussion in 'Events' started by Serge Pustelnik, Jun 14, 2005.

  1. May I also assume that because you have edited the IP addresses from the Records that Genesis routes its Clients orders over the Internet rather than over a Private dedicated Circuit that By-passes Publicly accessed Networks?

    If this is the case it imply that they, and more importantly their Clients, are at the mercy of whatever Internet Carrier they route their Orders over in order to reach the Exchange in Question.

    Please, if I am wrong, correct me. I can only conclude this is the case, otherwise there would be no need to Edit out IP addresses that run over a Private Network.

    I would appreciate it if you supplied a Copy of Traceroutes to the IP's in the first Post.
     
    #11     Jul 3, 2005
  2. EricP

    EricP

    Hmmm. So, we can assume that you did the same with your existing broker? Do you mind telling us who your existing broker is, and posting the above data that you received back from them? It would be nice for comparison.

    In fact, I would be interested in hearing the broker name, and associated data in your list, above, for ANY broker that provides this data for their firm. I have seen none that do. A web link would be great.

    Thanks,
    -Eric
     
    #12     Jul 4, 2005
  3. EXACTLY. This important "latency" stuff is critical to traders, but likely NOT PUBLISHED by brokers or order-entry middlemen. I recall that Barron's attempts to get a ranking on speed of execution, however, I'll bet their techniques are much less than scientific.
     
    #13     Jul 4, 2005
  4. nitro

    nitro

    Direct connections to the exchanges would not give you faster ping times than any of the above that are 1 to 2 MS. Those ping times are reflective of being on the same LAN!

    There is no way you are getting better ping times then 1-2 MS, even if your machine was sitting next to the matching engine on the switch!

    nitro
     
    #14     Jul 10, 2005
  5. hmap1

    hmap1

    "There is no way you are getting better ping times then 1-2 MS, even if your machine was sitting next to the matching engine on the switch!"


    I have to disagree with that statement. If you are pinging in Windows, yes, that is the best resolution you can get. Linux is a different story......You can get microseconds...
     
    #15     Jul 10, 2005
  6. hmap1

    hmap1

    For those who doubt: This ping is basically between two of my machines on the "same switch" as you have stated Nitro:
     
    #16     Jul 10, 2005
  7. nitro

    nitro

    Ah, No.

    While there are ping programs that do have microsecond resolution, that does not mean your connection is faster. That is like saying that I am going 5.50 miles per hour, but you are going 9 Kilometers per hour, therefore you are going faster. 5.50 MPH = 9 KMPH.

    You can rewrite the windows ping utility the same way to use the same strategy as microsecond ping resolutions programs and get the same answer.

    I think the reason the typical ping is written this way is so that it can interact with DNS servers. If you are doing point to point with no DNS, probably better to use other ping strategies.

    nitro
     
    #17     Jul 10, 2005
  8. hmap1

    hmap1

    I get resolutions smaller than 1 ms. I know my stamps, stats, execution times, hit rates, change in hit rates and everything else in real time and on a daily basis where I monitor it. Thats my game. I respect your difference in opinion. I know you have been at this game for years like I have.

    To state that you cannot get better ping time than 1ms is just plain wrong in my opinion.

    Best of luck.
     
    #18     Jul 10, 2005
  9. nitro

    nitro

    Well, you don't have microsecond resolutions, but that is ok. I am going to call Cisco Systems tomorrow to verify what I believe is true either way just in case I am missing something, but I really doubt it.

    Once the packet leaves your computer out the NIC, it is dominated by network physics. If I am on a LAN, there is no way to go faster than that. The only place where you could be gaining is in a faster TCP/IP stack by using a highly optimized Unix stack. But that would not go from milliseconds to microseconds.

    Best of luck to you as well.

    nitro
     
    #19     Jul 10, 2005
  10. hmap1

    hmap1

    My pings can get microsecond resolution using Linux. Nothing I do automated for speed is Windows, so I don't really care to change that metric. I performance monitor every app I run in realtime. My custom low-level IMDB runs at a certain CPU level and I monitor that during different mkt times with all feeds pumping through it full DOB. Good lines, locations and custom mods make a huge difference in performance. It's been trial and error over the years and when I started, we were not shaving times as we are now, but the game was different back then. :) Best of luck to you as well.
     
    #20     Jul 11, 2005