DAC Limits Change for End of Jan 06

stevie · Jan 20, 2006

From 31st January the limits will be increased to a maximum query rate of 1,000 queries per rolling 60 seconds and a maximum of 432,000 per rolling 24 hours, recalculated every five minutes.

http://www.nominet.org.uk/other/dac/aup/

FC Domains · Jan 20, 2006

Why?

Everybody will just run their systems faster and put a bigger load on Nominets servers.

retired_member11 · Jan 20, 2006

The schizophrenia of Nominet.

MALACHI · Jan 20, 2006

FC Domains said:
Why?

Everybody will just run their systems faster and put a bigger load on Nominets servers.

They know how much verisign is making and want a bigger slice of the pie

fred · Jan 20, 2006

MALACHI said:
They know how much verisign is making and want a bigger slice of the pie

A don't see how Nominet increasing the DAC usage limits would get them "a bigger slice of the pie"

grantw · Jan 20, 2006

I'd say this idea comes straight from the Nominet employees handbook:

http://www.amazon.com/gp/product/0446393592/ref=nosim?tag=acorn06-21

Seriously though, if it doesnt overload the system then why not? Its not gonna make any difference to who gets what as everyone is just gonna double the speed of their systems.

Grant

retired_member11 · Jan 20, 2006

So change for the sake of change then? If this change means relatively speaking no change at all in terms of who gets what then why bother?

My Ferrari can do 200 mph... but not in the rush hour. Bad analogy but do you get my point? Unless this analogy hints at allowing 'faster' systems to outpace slower systems to an even greater degree?

retired_member9 · Jan 20, 2006

Nominet, if they would have been just a bit bothered about the integrity of their AUP, rather would made limits less favourable for those who decide to take advantage of having multiple (friends, colleagues) Tags.

The formula is easy:

Code:

r - number of allowed requests per minute
t - number of allowed requests per 24h

r * 1440 <= t

The higher is 'r' the better.
It would be *ideal* to have r = 4000 and 't' according to the formula.

Alas, Nominet servers most likely couldn't stand that

aqls · Jan 20, 2006

If Nominet suddenly has this 'extra capacity' then it should also increase the number of speculative requests.

After all, speculative requests are a fairer market and the only previous explanations from Nominet about not allowing more of them have been for technical reasons.

-aqls-

retired_member26 · Jan 20, 2006

invincible said:
Speculatives require more resources, we have been told. I don't ever expect them to increase the limit. EPP will probably change things. Why do you think speculatives are fairer? Do you mean "they don't require the operator to be as technically minded"?

No,

We think that if you increase the limit of speculatives it means that you completely erase the possibility (underlying the word possibility, if it exists) of fraud of having double/triple/quadraple+ TAG/membership bringing consequtive double/triple/quadraple+ DAC quotas.

Again it is a possibility that you could have a TAG for as A. RIGHT and have a TAG on your uncle as W. RIGHT and use them DAC quotas. However, if you increase spec quota every TAG must send a spec with its header.

Besides spec emails are as technically minded as dac.

TurNIC

FC Domains · Jan 20, 2006

turnic said:
We think that if you increase the limit of speculatives it means that you completely erase the possibility (underlying the word possibility, if it exists) of fraud of having double/triple/quadraple+ TAG/membership bringing consequtive double/triple/quadraple+ DAC quotas.

Again it is a possibility that you could have a TAG for as A. RIGHT and have a TAG on your uncle as W. RIGHT and use them DAC quotas. However, if you increase spec quota every TAG must send a spec with its header.

Besides spec emails are as technically minded as dac.

The way I read that, it's total bollocks. (I think I hear the distant thudding of Invincible banging his head on the table.)

Multiple identities would give you an advantage under any system. DAC, speculative, whatever.
DAC is more technically demanding.

paul · Jan 21, 2006

grantw said:
Its not gonna make any difference to who gets what as everyone is just gonna double the speed of their systems.

Grant

That's not how I see it.

If catcher A has a system that responds in say 50ms and catcher B has one that responds in 1ms then catcher A still has a chance of catching something if he hits the DAC at just the right time. As catcher B can only query every 120ms

so in theory catcher A could get a response 119ms earlier than catcher B if he was very lucky.

Let's say the domain internet.co.uk drops at 01:10:00,000 and both catchers are querying at 500 per minute evenly spaced (every 120ms)

Catcher A's request hits the DAC at 01:10:00,000 (really lucky)
Catcher B's requests hit the DAC at 01:09:59,999 & 01:10:00,119 (probably me)

Catcher A responds in 50ms = 01:10:00,050
Catcher B responds in 1ms = 01:10:00,120

Giving cather A a 70ms window of opportunity

Now if they can both query twice as fast that window would be reduced to just 10ms

A faster query rate gives those with fast systems an increased chance of catching names.

Paul

Local Shops Adult DVDs HPI Check Domains

retired_member11 · Jan 21, 2006

I didn't do the numbers but this analysis, if correct, confirms my earlier intuitive feeling that the increase in query limits would be to the advantage of the faster systems. I've no doubt that these faster systems have over the years developed an 'affinity' with the Nominet technocrats with 'suggestions' working their way into the architecture of the process. The whole thing needs stripping down. I still favour Nominet authoring their own script package and distributing it to TAg holders under licence.

paul · Jan 21, 2006

argonaut said:
I still favour Nominet authoring their own script package and distributing it to TAg holders under licence.

Where's the fun in that

The challenge is getting your system to perform more like catcher B than catcher A

It's how it should be, the ones who have invested the most money and time into their systems get the rewards.

Paul

Local Shops Adult DVDs HPI Check Domains

tifosi · Jan 21, 2006

argonaut said:
I didn't do the numbers but this analysis, if correct, confirms my earlier intuitive feeling that the increase in query limits would be to the advantage of the faster systems. I've no doubt that these faster systems have over the years developed an 'affinity' with the Nominet technocrats with 'suggestions' working their way into the architecture of the process. The whole thing needs stripping down. I still favour Nominet authoring their own script package and distributing it to TAg holders under licence.

sure, sit on your arse all day and let others do the hard work. pah!

tifosi · Jan 22, 2006

grrr

I know it's only semantics, but it's late, and I'm grumpy...

MEMBERSHIPS query the DAC not TAGS.

Got that folks...

right. Only semantics, but crucial to the conversation.

aqls · Jan 23, 2006

So more and more:
there will be a cut off point for slow systems
a) your system will have to be at least x ms in order to even be in the game for the best names
b) systems that stick rigidly to the Nominet preferred guidelines or systems that don't use 'tricks' (router manipulation, packet level manipulation, short personal and address details etc) will be out of the game
c) TAGS that don't have many business relationships with 10 or more other members to use their ip address will be out of the game
d) I would be surprised if noone has cracked their systems yet anyhow.

It is only partly techie wins all, it is more streetwise savvy wins all.

For the most of us on this board, it means we're left fighting over scraps. Whichever way you look at it, it means more domains into less peoples hands.

If your good at that sort of fighting - best of luck.

I hear many on this board saying they are catching less and less.

Time for a more fair system.

More speculatives and Nominet providing the software.

I am not calling for a random distribution of names, but a system that isn't heading headlong into a ridiculous state.

-aqls-

retired_member11 · Jan 23, 2006

aqls said:
So more and more:
there will be a cut off point for slow systems
a) your system will have to be at least x ms in order to even be in the game for the best names
b) systems that stick rigidly to the Nominet preferred guidelines or systems that don't use 'tricks' (router manipulation, packet level manipulation, short personal and address details etc) will be out of the game
c) TAGS that don't have many business relationships with 10 or more other members to use their ip address will be out of the game
d) I would be surprised if noone has cracked their systems yet anyhow.

It is only partly techie wins all, it is more streetwise savvy wins all.

For the most of us on this board, it means we're left fighting over scraps. Whichever way you look at it, it means more domains into less peoples hands.

If your good at that sort of fighting - best of luck.

I hear many on this board saying they are catching less and less.

Time for a more fair system.

More speculatives and Nominet providing the software.

I am not calling for a random distribution of names, but a system that isn't heading headlong into a ridiculous state.

-aqls-

"...and Nominet providing the software." Nominet should have done this a long time ago.

tifosi · Jan 23, 2006

It's fine calling for Nominet to provide the software. But you have to consider that:-
1. It's machine independent - there's a world of difference between a dual xeon linux server running fedora, debian, or whatever, and a desktop pc running windows xp home.
2. Language dependent - ie it will have to run on windows pc, windows servers, *nix servers, and come in multiple flavours of programs i.e php, asp.net, c#, bash.

And how are you going to validate how this is going to work if essentially all it does is open a socket, call, receive a response and evaluate it. The 'program' would have to be such that it could not be decompiled and 'bettered', and have a signature that could not be transferred to another mimicking program.

There needs to be some hint of realism here. I'm all for a level playing field, but this has to come from there geniuinely being no behind the scenes manipulation of the system.

This is synonymous with sport providing the doping agencies and out-of-competition testing, not by giving the athletes the same diet and watching while they spoon it in.

Stephen

retired_member9 · Jan 23, 2006

Oh, common, it is such a silly demand from Nominet.

They simply delete unpaid domains, why would they care about those who
want to catch them??

The one and only thing I can see they might do in favor of dropcatchers peace of mind, is to disclosure their "random" deleting algo, as it should be clean enough to prevent "pattern recognition" if it exists.

Good random algorithm can't be weakened by examining & studying it.

But if it's not -- people who are behind it have a "dangerous" knowledge.