November 14th, 2008


I'm a mummy!?! SQUEEE!!!!!

This news is massively overdue as I've been busy with a new job and all the other chaos that follows the arrival of a baby, but I would like to formally announce the birth of Thomas Peter Gearoid Loughlin-McHugh, born 23:02 on 7th October 2009 at a whopping 9lb 2 oz. He's absolutely gorgeous (photos to follow later on my flickr stream).

goth_twiglet and little Thomas are both doing well and the menagerie all seem to have forgiven us for acting in an inexplicably doting manner!
deep geek

dotTel API announced

Telnic's dev site now includes details of their JSON API for provisioning and manageing dotTel domains. Looking though the examples this morning it should be easy to roll up a Ruby engine for managing provisioning and as landrush release isn't due until early February that will give spikyblackcat and I lots of time to devise something interesting.

There's also the possibility that I'll be doing another public presentation next month on Push using EventMachine, and if I can find a way of spinning in a dotTel component to that I will.

Those of you who know the tempestuous relationship that spikyblackcat and I had with Telnic will probably be surprised that we're still following this project and attempting to get the technology to a wider audience, but egos and personalities aside this stuff is just so damn addictive once you get into it that we would both love to see dotTel succeed.

I know DNS is considered both an obscure black art and a bit of a backwater, but dotTel along with ENUM are the first of a new generation of DNS techniques that will make it indispensible to software developers deploying large-scale distributed applications. Admittedly there are some issues with using DNS at the boundary of its dynamic behaviour (not least the relative inefficiency of caching with a 60s or lower TTL) but the same's true with media serving and many other applications which have far heavier bandwidth and transaction costs.

The other shakeup is going to be in forcing the pace of adoption of ENDS0 as a reasonable subset of dotTel zones will require TCP connections due to their packet size. This could be the start of a move to a more persistent relationship between clients and DNS servers, and given that DNS traffic needs to be fairly efficient it may even open the way to equivalent protocols based on RUDP. spikyblackcat and I have been considering these issues for a couple of years now and working on solutions but it's only going to be when an identifiable demand is established that people will finally get what we're talking about and why.

Anyway, here's to a successful launch of dotTel!