Enjoy unlimited access to all forum features for FREE! Optional upgrade available for extra perks.

Removed catching thread.

Discussion in 'General Board' started by PoshTiger, Oct 23, 2014.

Thread Status:
Not open for further replies.
  1. AssetDomains

    AssetDomains Well-Known Member

    Joined:
    Feb 2010
    Posts:
    3,062
    Likes Received:
    76

    Given that they catch anytime in a 24 hour period and you agreed that drop times were random as far as we all know then there scripts must be polling 5 times a second or 200ms

    What was wrong with that assumption.
     
  2. Domain Forum

    Acorn Domains Elite Member

    Joined:
    1999
    Messages:
    Many
    Likes Received:
    Lots
    IWA Meetup
     
  3. ian

    ian Well-Known Member

    Joined:
    Jan 2008
    Posts:
    4,154
    Likes Received:
    311
    This all helps, but there was a very obvious change in success when a few domainers started catching everything, so unless linked, I cannot see them all just so happening to invest the vast amounts of money in the setup you've mentioned at the same time. It seemed to centre around a Nominet update, but maybe that was just a coincidence.
     
  4. ian

    ian Well-Known Member

    Joined:
    Jan 2008
    Posts:
    4,154
    Likes Received:
    311
    Based on catch times I've recorded, I would agree this is likely, but why still such an advantage. I've been beaten by presumably 5/sec whilst querying at 32/sec.
     
  5. invincible

    invincible Well-Known Member

    Joined:
    Feb 2005
    Posts:
    4,203
    Likes Received:
    101
    I was asking why do you feel you need just 200 extra DACs to cover every 1 of the milliseconds in that 200ms gap? Why not break that 200ms into picoseconds and then claim you'd need even more DACS (more than 200) to cover each picosecond in that 200ms?

    I don't believe one would need so many extra DACs. I do believe that more DAC resources distributed across the day would give exponentially better return at least to a point. So having 10 will be better than having 2 or 3. We almost appear to be forgetting how brief a millisecond is. If you've got decent coverage on the maximum speed setting across the day, which can be achieved with 4 tags, and perhaps a few more, you will have better coverage than 1 tag could achieve.
     
  6. ian

    ian Well-Known Member

    Joined:
    Jan 2008
    Posts:
    4,154
    Likes Received:
    311
    This is true, but no greater advantage over any other catcher that just so happens to be querying at the same 16/sec speed at that same time. You would need a lot more tags to counter that, and the rest.
     
  7. invincible

    invincible Well-Known Member

    Joined:
    Feb 2005
    Posts:
    4,203
    Likes Received:
    101
    You presume. It'd depend on how many other members were querying the DAC at the time the domain name dropped. If you can blanket cover the entire day on full speed, overlapping with extra DAC connections at certain times, and you're lucky that enough good domain names drop when you've got good coverage compared to the sum of the other members operating at the same time as the drop, plus your connectivity, server and code are tip top, I feel you'll get a better outcome than if you are playing with the one membership.

    What drop catchers usually forget is they have no way of knowing how many memberships were operating and at what speed at the exact time the domain name they caught actually dropped. Therefore it's hard to know how much better or worse the successful drop catcher was than anyone else.
     
    Last edited: Oct 23, 2014
  8. AssetDomains

    AssetDomains Well-Known Member

    Joined:
    Feb 2010
    Posts:
    3,062
    Likes Received:
    76
    Because I assumed there was approximately 100 TAGS all polling at the same time each going for a name 5 times a second all being equal which I know its not.
    From that 500 requests for the domain per second or one request coming in every 2ms on average.

    If supper catcher A got a not available response then in the time between poll's there should have been 100 requests all checking if the name is available if there systems were even half decent they should have been able to register the name long before the 200ms were up my home IP can get a response in 10ms

    Looking back through my old logs when I wasn't hosted by anyone 100 does seem a bit high now probably more like 30 - 40
     
  9. Retired_Member42

    Retired_Member42 Retired Member

    Joined:
    Sep 2009
    Posts:
    1,199
    Likes Received:
    22
    Even on full speed you're still going to lose out. Sure you'll know the domain has dropped quicker if you're running at 60ms than the 200ms guy but what if that 200ms guy queried 140ms before you and their script is slightly faster? What if your script took 140ms more to send the reg request than the other guy? Getting the notification, sending the reg request and Nominet processing it takes a lot longer than an availability poll.

    If I had 4 memberships/tags I'd be looking to stagger them all at 200ms for the full 24 hours. Not simply covering the full 24 hours at 60ms. I feel you'd have a much better chance at catching this way.
     
  10. ian

    ian Well-Known Member

    Joined:
    Jan 2008
    Posts:
    4,154
    Likes Received:
    311
    Good point, instead of 16/sec on 4 tags (with obviously an overlap in timing somewhere), you'd be running at 60ms, where as 4 tags split evenly at 5/sec all day/night, you'd be running the equivalent of 50ms.
     
  11. martin-s United Kingdom

    martin-s Well-Known Member

    Joined:
    Jul 2012
    Posts:
    3,468
    Likes Received:
    270
    I haven't bothered to test it, but I also wonder whether it's not just the interval, but the timing of that interval.

    Maybe checking at 0.2, 0.4, 0.6, 0.8, 1.0 isn't as effective as 0.22, 0.42, 0.62, 0.82, 1.02.

    I believe you can calculate the exact time on Nominet servers based on knowing the latency of a request/response and querying one of the timestamped EPP records?

    I've had lots of thoughts about optimisations, but not tried many of them out recently.

    For the record, I'd agree with the observation about response times pre-drop.
     
  12. Retired_Member42

    Retired_Member42 Retired Member

    Joined:
    Sep 2009
    Posts:
    1,199
    Likes Received:
    22
    There could well be a pattern. Even domains dropping in the last half of a second would double your coverage / amounts of queries within a given window. If domains are deleted from Nominet at a fixed time at the start of a whole second for example and they take X mliliseconds before coming available - if you knew that figure, that in itself would be a massive advantage.

    But even knowing that, I still think a slick script will win over polling speed every time. Even if you can manage to 2, 3 ,4x your polling speed.
     
  13. SF

    SF Well-Known Member

    Joined:
    Jun 2005
    Posts:
    7,543
    Likes Received:
    58
    Who is this order would you put at the top tags

    1:
    2:
    3:
    4:
    5:
     
  14. invincible

    invincible Well-Known Member

    Joined:
    Feb 2005
    Posts:
    4,203
    Likes Received:
    101

    I'd say you have the tiniest one by far.


    (from iPhone)
     
  15. SF

    SF Well-Known Member

    Joined:
    Jun 2005
    Posts:
    7,543
    Likes Received:
    58
    so i am 5th that ill do nicely?
     
  16. bonusmedia

    bonusmedia Well-Known Member

    Joined:
    Oct 2012
    Posts:
    1,411
    Likes Received:
    156
    Sorry to barge into the technical talk...

    What is the ROI on catching?

    What do you actually earn per hour of research and setup etc, assuming you rent a script at say £50 per month and you already have a Nom tag?

    If I finally get a Nom tag, am I actually allowed to lease out my DAC, and if so what's the going rate?

    Is there some way of finding the real premium drops without trawling through droplists for hours?
     
  17. invincible

    invincible Well-Known Member

    Joined:
    Feb 2005
    Posts:
    4,203
    Likes Received:
    101
    Depends how much you sell what you catch for. I think it can be a nice little earner a month but probably not something one should solely rely on.

    No idea what people make nowadays. If you're catching to keep rather than catching to flip quickly via end user or DL you won't make any money immediately. Many of today's catchers seem to want to flip their caught domain names much more quickly. That could either be because margins are now thin and they cannot afford to keep the names long or because multiple catchers are pooling resources hence are forced to sell so everyone contributing to the resource pool gets their cut of cash.

    Best you ask Nominet. :)

    The daily lists shouldn't take you long to scour. It's the juicy out of sequence suspended domain names that some catchers go for that occur for a variety of reasons, often because someone notices a ltd has been dissolved and reports the associated domain name to Nominet. Once the domain name has been suspended some catchers have an idea of how long it will be before it drops. I also believe Nominet have sometimes given away the future drop date of out of sequence suspended domain names, or at least the date the domain name was suspended. That might depend how pally you are with someone in Nominet CS.

    I think it could be easy to burn £10k on an ultra high speed setup. I don't believe any catchers are effectively their own ISPs and lonap members, hence they don't peer with Nominet directly, but twelve months of colo space, setup costs of running cables, purchasing servers, a router, a switch lonap setup and fees could wipe out that kind of budget. However most catchers seem to simply lease server space at one of a number of ISPs that they believe give them good connectivity to Nominet (via Linx or lonap), perhaps because that ISP will plug them into the switch closest to their Linx or lonap router port.

    Aside from this operating multiple memberships will give you an advantage over trying to work with one. If it didn't, why do Nominet ban it in their rules? The problem with multiple memberships is unless you're lucky enough to have someone who is happy to let you use it without any expectation of a return, you'll have to compensate them some how. That can be financially or with catches or with a cut of sales. Trust me, pooling of resources and splitting of sales prices of the resulting catches occurs. :)


    (from iPad - K)
     
    • Like Like x 1
  18. WalkinDude United States

    WalkinDude Well-Known Member

    Joined:
    Jul 2012
    Posts:
    1,140
    Likes Received:
    95
    Another fantastic row. Gripping read. Well done to all participants. This place is getting more interesting than TheGuardian. Not being sarcastic either.
     
  19. Retired_Member39

    Retired_Member39 Retired Member

    Joined:
    Sep 2013
    Posts:
    569
    Likes Received:
    8
    I think this thread has been a decent civil discussion. Some very good points from the likes of Invincible about the factors involved in certain people being more successful at catching domains which has helped me understand more about how it all works. Thing is if this isn't properly explained and it's just oh x has a better script then ofc questions might be asked if person x and y are catching everything. To the average Joe the main thing was just how could someone with 1 DAC quota beat 100 DAC quota's consistently.

    I think it's perfectly reasonable for the dozens of members who have spent x,xxx's on nominet membership/catch credits/hosting chasing to ask questions in a reasoned manor.
     
  20. julian United Kingdom

    julian Banned

    Joined:
    Aug 2007
    Posts:
    2,252
    Likes Received:
    41
    the meat of it is, if Nominet finds there has been unscrupulous behaviour by a tag holder, what sort of punishment are we looking at?
     
  21. Retired_Member39

    Retired_Member39 Retired Member

    Joined:
    Sep 2013
    Posts:
    569
    Likes Received:
    8
    An evening with Invincible.
     
Thread Status:
Not open for further replies.