There is an interesting BusinessWeek article by Olga Kharif mentioning our efforts with Android. Although our work is in the very early stages and we have yet to see how successful Android will be with major operators and handset providers - article gives good overview of the current state of the art and provides some thoughtful prognosis. Good read overall.

I do believe that mobile grid computing would be one of the next breakthrough enabling technologies for mobile market. Stay tuned!

I'm going to be presenting at JBoss World 2008 in Orlando on February 14th on topic of integration between GridGain and JBoss. If you are around - come see my presentation!

With upcoming GridGain 2.0 release we are going to round up our integration plans with JBoss by providing what I consider truly native blend-in integration. In fact, when you use GridGain with JBoss, GridGain basically becomes a part of JBoss container that looks as if it just acquired new functionality. Here's the list of integration points that we implement in totality with GridGain 2.0 (most of them has been available since GridGain 1.6):

• Loader for JBoss AS
• Support for and integration with JBoss AOP
• Integration with JBoss JMX
• Integration with JBoss Logging
• Integration with JBoss HA
• Integration with JBoss Cache (including support for dynamic data partitioning, load balancing and affinity map/reduce)
• Integration with JGroups

The reason I'm really excited about this integration is that it provides the first fully open source full-stack grid computing platform for Java, combining state of the art compute and data grid functionality. And that's why I think open source Java grid computing has never been more exciting!

Almost everyone who comments on the idea of mobile grid computing puts forward the skepticism about its possibility due to... battery performance in today's mobile devices.

So, the usual thought goes that if I let my cell phone crunch some numbers for extended period of time, let's say an hour, it will drain battery dry in the same hour. So, who needs it, right?

No. What's everyone is missing that mobile grid computing will require a slightly more sophisticated Map/Reduce implementation, specifically sparse load-balancing. Sparse load-balancing algorithm will perform splitting in such a way that any individual mobile device will only participate in the grid for short periods of time significantly reducing battery drainage. This logic is permitted by a simple fact that you may have tens or hundreds of thousands consumer mobiles devices in the grid at any given point of time - and you can use sparse load-balancing on such grid.

Sparse load-balancing is one of the areas in which mobile grid computing is fundamentally different from traditional server-based grid computing where utilization is the key. In mobile grid computing we are actually trying to reduce the utilization of an individual device but rely on a sparse distribution.

Another area where mobile grid computing is special is the use of redundant split, i.e. moving the same job to more than one node for execution - again relying on the fact that there are plenty of available grid nodes but each node has limited "reliability" for finishing its job.

Combination of sparse load-balancing and redundant split on mobile grid provides guaranteed execution with minimal impact on an individual mobile device.

With LEGO-like customization of every aspect of grid infrastructure GridGain is a perfect technology for mobile grid computing. Topology management and discovery, load balancing and dynamic map/reduce are all fully pluggable and customizable in GridGain.