Until now, the internet has largely been used as a tool for communication – but new applications are starting to use the internet to coordinate work between machines. This could lay the basis for a whole new level of automation across society, leading to great gains in productivity. The first step in this process is ‘grid computing‘.
This article is a good decription of grid computing or what others may call High Performance Clustering or Distributed Processing. When implementing HPC/DC, the biggest problem that needs to be overcome is the idea that the tasks that you need to be worked on must be designed & implemented in a parallel fashion. Grid computing is wonderful on free OSs like Linux or BSDs but most of the time (not process migration like OpenMosix) require that the program to be executed to be specially complied against parallel libraries. Common PC user tasks will not benefit from grid computing. I don’t care how many FLOPS or MIPS a cluster can generate if I need to read a webpage and understand the material or if I need to write an email to a friend.
“But the vision is that they will one day connect together through the internet into a single global grid, so that the combined computing power of the entire world will be available from every desktop.”
While I aplaud the author for being a visionary (he sounds like an evangelist), I know that at least for now that goal is aways off. I know some companies need to do processing offline or offsite so that it gets done, but connecting desktops from around the globe to do what exactly?? Payroll processing? Theoretical protein synthesis? This model of computing will not be easily embraced by the populous. Companies with special needs. Yeah sure. Common people? Probably not.
*EDIT*
I don’t care how many FLOPS or MIPS a cluster can generate if I need to read a webpage and understand the material or if I need to write an email to a friend, blazing fast speeds on a cluster across a distance will not help at all.
“You will never need more than (8k, 8M, 128M, …) of memory.”
Just because you can’t see a reason for it now, does not mean that it should not be done, or will never happen.
Sounds like more dotcom hype to me. Which companies are paying this guy?
–ms
there is no need for a home user to use grid computing…grid computing is for number crunching. perhpas one day when we have a global society a massive grid will be used to store, compute, and coordinate all information and systems…that is a very good idea to base all the worlds infrastructure off of a grid computer becasue of the fault tolerance.
Say you are a home user doing Video processing or 3D Cad / Graphic Designs? How about Compiling complex tasks that you are doing? These are complex tasks that require heavy duty processing. They can also be modified to be done in Parallel tasking (read, one huge render farm) These are some of the uses that many home users can use today.
How about using the GRID network to process SETI information on extraterrestial life? wait…. that’s already being done.
You guys are so pessimistic you can’t see the uses.
In a broad and academic sense, yes grids will become raw computing power. But on the commercial side of the grid, this is more than just “compute” cycles – it’s the ability to take raw compute cycles and virtual storage to create applications that live on the the grid. For example, a mostly hosted version of MS-Office or the Slashdot.org website or an instance of VMWare. In the world of the grid, your programs would be totally virtualized on the grid. The benefit to the big computer companies is that they can take their products and turn them into services (with term contracts). This gives them more predictable revenue streams. The benefits to the companies that consume grid based services is a) leased access to software – hence lower overhead or commitment and b) better integration (since most of this functionality will be web services – or so they say). To the home user, who knows what this will mean – probably a good bend over the table!