Welcome to our blog. Here you will find postings by the Parabon team on a variety of topics relating to grid computing. But, don't be surprised if you find occasional musings on chaotic systems, econometrics, structural DNA, model rocketry, the fluid dynamics of beer suds or any of our other passions that are somehow related to our unifying desire to compute deeply and help others do the same.
Distributed Computation — A Primer
By Dabe — 28 July 2008
What follows is an attempt to describe distributed (grid) computing in layman's terms. We'll take a simple problem and analyze how to go about solving it, avoiding some common pitfalls along the way.
So let's say we have a Job. For this example, we're going to figure out which name in the phone book has the highest Scrabble score. For the purposes of this demonstration, let's assume that it takes one second to calculate the score for one name, and that there are a million names in our phone book. Thus, one computer working end-to-end would take one million seconds, which is 277 hours, or roughly 11.5 days. Obviously we can do better by splitting up the Job into multiple smaller tasks. [But being as this is a Distributed Computation primer, I'm guessing you already knew that!] (Read more...) Throw Millions of Dice — and Ignore Half
By Sean — 22 May 2008
When you first take your embarrassingly parallel, embarrassingly slow application and throw it on a big grid, it can sing and dance and return results fast enough to make you not really wonder — nor particularly care — about how it got from here to there. Indeed, when your grid is built on top of unreliable and inhomogeneous resources, you probably really don't want to think too much about how the sausage gets made — and a well-built grid, such as Frontier, insulates you from such concerns for the most part by being intelligent and playing games with the resources on your behalf so that they behave like a team of workhorses rather than the willful herd of cats they are. But once you work with a grid for a while and start to get familiar with its behaviour and quirks, then thinking a bit about what's actually going on behind the scenes can help take an app that goes and make it hum instead. Case in point: Monte Carlo.
But first, let me explain what I mean by the "unreliable and inhomogeneous resources" composing the grid. (Read more...)
Wall Socket Accessibility
By Steve — 1 May 2008
It has been nearly a decade since we founded Parabon and introduced the commercial market to grid computing. Although our marketing firm at the time successfully advised us to avoid choosing an "e-name" for the company, we nevertheless could not resist all of the Net forces of the day, which is why we originally described our service as Internet computing. After the dotcom crash revised the Recommended Daily Allowance of "e" and "Internet" language, I was pleased to see the widespread adoption of the term grid computing because of the strong parallels between the electricity grid and the computation grid. Although the marketing machinery at various "Big Iron" hardware vendors has somewhat diluted the definition of grid computing to mean practically any distributed computing solution that promotes the sale of more computers, to us it still conveys what we have always aimed to provide — a high-performance computation (HPC) utility that is as easy to access as electricity from a wall socket. (Read more...)
