I wish someone would give me a column even though I don't know what the fuck I'm talking about. Greenspan had a role in the current financial crisis, but he was not the sole engineer, nor even the biggest one. For actual, sound info on this subject, check out the EconTalk podcast/blog.

I thought getting a column required that you know nothing ;p well on the internets anyway

History will judge this to be our day's version of VBA.

If you want a framework with scalability "baked in", so to speak, you should take a look at GigaSpaces XAP (http://www.gigaspaces.com/). Its not the same model as the others mentioned on this thread but will prevent that "big rewrite" in the event that you do indeed become successful enough to need it.

Clojure is neat and all, but I can't see it being anything but a proving ground for some more esoteric technologies or methods. Lisp has been around too long to get mainstream acceptance and the "power vaccum" of which PG is fond of mentioning is marginal these days due to Ruby, Python, etc. Clojure has the attractiveness of running people's existing infrastructure (i.e. the JVM) but its nowhere near being alone in that regard. Its neat but don't look for it to be a world-changer.

I'd be interested to see concurrency baked into Ruby or Python the way it is with Erlang and Clojure. Until then, my eye is on Clojure.

Joel, this will be an attempt to prove out your OpenPoker infrastructure to either a) sell licences to OpenPoker, or b) sell the rights to it outright, correct? Given that, the site does not need to be super-popular in and of itself. You will probably only need about 100,000 active players to prove how much more efficient and scalable your software is. My thoughts on making money are MochiAds and private tables; they seem about as good as any. But if the goal is to sell OpenPoker, then you just need a proofpoint, not a successful site in and of itself (in fact, a popular site would make it harder to sell off the rights to OpenPoker altogether, as it would raise the cost to a potential acquirer).

Google won't acquire GitHub because a) they're not for sale just now, and b) Google has a history of avoiding companies that don't share at least some of their underlying technologies. GitHub's Rails/EngineYard combo is probably enough to put GOOG's technical M&A team off the track. Plus, they would have no interest in Git internally, having built a ton of cool tools around their version of Perforce.

> Perforce

yeah but google code doesn't use perforce either... so it's possible (not saying likely). I tend to agree with the rails thing...

Digital Element has the best data, but they are on the expensive end and their software isn't all that flexible in terms of deployment. Maxmind GeoIP is weaker in terms of accuracy but are far more flexible.

You should have probably read more closely before writin this diatribe: they specifically note in several places that they do not use TCP and can also max out on 4x Infiniband as well as 10GigE. They use UDP as their transport for ZeroMQ messages.

I dimed this out and had it disabled. Twitter doesn't need people poking through its' SQL. Imagine what we'd find if we were poking through yours...

We saw this today and have already switched over to it. Literally three hours later we are using Graphite for our health and performance graphing needs. It took about 14 lines of Python to hook up to our existing stuff. Awesome.