Comments on: Gigaspaces launches eXtreme Application Platform http://insidehpc.com/2007/06/12/gigaspaces-launches-extreme-application-platform/ HPC News Without the Noise for Supercomputing Professionals | insideHPC Sun, 19 May 2013 10:46:54 +0000 hourly 1 http://wordpress.org/?v=3.1.1 By: Geva Perry http://insidehpc.com/2007/06/12/gigaspaces-launches-extreme-application-platform/#comment-2509 Geva Perry Sun, 17 Jun 2007 18:46:14 +0000 http://insidehpc.com/2007/06/12/gigaspaces-launches-extreme-application-platform/#comment-2509 John -- We would be happy to schedule a briefing with you to talk about GigaSpaces. Please send me an email if you would like to do so. Geva Perry GigaSpaces John — We would be happy to schedule a briefing with you to talk about GigaSpaces. Please send me an email if you would like to do so.

Geva Perry
GigaSpaces

]]> By: David Teszler http://insidehpc.com/2007/06/12/gigaspaces-launches-extreme-application-platform/#comment-2415 David Teszler Thu, 14 Jun 2007 21:11:15 +0000 http://insidehpc.com/2007/06/12/gigaspaces-launches-extreme-application-platform/#comment-2415 Let's see if I can't get this started. The Application pattern will make the call of value for these frameworks. Gigaspaces, based on the open source Javaspaces reference implementation, has what one might call a balanced software architecture. Compute throughput is captured in the notion above of 'writing to a single computer and then scale out the application on multiple processors.' The "balance" enters when the application grid is combined with the data grid. In many scenarios in HPC, getting data to the application is the bottleneck. Data Grid addresses the I/O component within it context. Orchestrating an application grid and a data grid allows jobs and data to be intelligently located. The new bindings appear to extend this to other languages allowing non-grid application access to the capability. Some might achieve a similar end via DRMAA. However, that still leaves the data i/o issues open. Of course the value of this depends on your problem. Let’s see if I can’t get this started. The Application pattern will make the call of value for these frameworks. Gigaspaces, based on the open source Javaspaces reference implementation, has what one might call a balanced software architecture.

Compute throughput is captured in the notion above of ‘writing to a single computer and then scale out the application on multiple processors.’ The “balance” enters when the application grid is combined with the data grid. In many scenarios in HPC, getting data to the application is the bottleneck. Data Grid addresses the I/O component within it context. Orchestrating an application grid and a data grid allows jobs and data to be intelligently located.

The new bindings appear to extend this to other languages allowing non-grid application access to the capability. Some might achieve a similar end via DRMAA. However, that still leaves the data i/o issues open.

Of course the value of this depends on your problem.

]]>