Tuesday, November 9, 2010

A thought experiment

I remember back in the day when companies needed to raise millions of dollars to set up their web farm. One of the upsides of doing that was that people thought big, they dreamed of taking over the world. These days of running a clustered multi-tier web app stack on AWS for pocket change has lead to people having small dreams.

So, how about we do some maths. Let's say you have $10k a day in budget to burn on compute resources. What could you do with it? Let's take numbers from AWS, they quote Linux Double Extra Large High Memory instances at $1 an hour. So, at 24 hours a day, we have about 416 instances. That is approximately 13TB of RAM, and over 1.6k cores. Now what can you do with that much RAM and CPU?

Thursday, October 28, 2010

Cool preso

A very cool analysis of a website's copy. Worthy of a detailed read.

Sunday, March 14, 2010

Mmm, stuff.

It's been almost a year since I last posted here. In the mean time I've wandered a bit, I spent a bunch of time playing with Google Wave, toying with Google Buzz, and poking at the corpse of LiveJournal. What brought me back? The fact that blogger has a new template editing system that doesn't suck nearly as badly as it did. In fact, it's kinda decent. About the only thing i want in it is a "round the corners and add subtle shadows" button. And maybe a Web2.0 Always in Beta sticker.

In some ways, this feels like i've been full circle. I'm back to writing Antlr grammars. Antlr 3.2's tree filters are wickedly cool, and Terence Parr's Language Implementation Patterns is thoroughly hurting my noodle. It's a damn fine read, if you enjoy LL(*) and syntactic predicates.

One of the things I am playing with at the moment is stuffing Antlr generated parsers inside of web browsers using Google's GWT. It opens a wild vista of new potentials, of being able to edit and hack seriously complicated models inside a browser. Revenge of the nerds, web2.0 style.