IBM announces Grid Computing for Linux

By Alun Williams

Posted on 1 Apr 2003 at 12:12

IBM and various Linux partners have announced toolkits and middleware to support Grid Computing on IBM-based Linux systems.

Specifically, Grid Computing will be available for enterprises running Linux on the IBM eServer zSeries mainframe. This is achieved using the Globus Toolkit for Linux which is part of SuSE Linux Enterprise.

'With this announcement, IBM is enabling a powerful Linux platform to participate in grids,' said Holger Dryoff, general manager of SuSE Linux. 'By enabling the world's most stable platform as the control unit for grid computing, there is the potential for a great leap in computing power.'

Grid middleware provider DataSynapse and Grid software specialists Platform Computing have also announced products to support clustering and workload balancing on the zSeries.

For example, DataSynapse's LiveCluster 3G is designed to avoid processing bottlenecks by better managing the power of servers and desktops anywhere on a network. Platform Computing's JobScheduler is designed to grid-enable applications and process flows across distributed computing clusters.

The goal of Grid Computing is to provide on-demand access to computing resources, on a high-availability basis. The significance of these announcements is that the concept and theory is being implemented in the real world. It represents a model for computing that possibly anticipates the massively networked systems of the future.

These developments follow on from IBM's recent efforts in the area of 'autonomic computing', which is where a computer system better manages its own operation (sometimes also referred to as self-healing software) - IBM breathes life into autonomic computing.

The three technologies announced at the start of March were 'Adaptive Forecasting', 'On-line Capacity Planning' and 'Rapid Reconfiguration'. First of all, mathematical models are used to forecast surges in demand. Then the system estimates the resources required to meet the now-expected levels of demand. Finally, Rapid Reconfiguration, manages a 'hot swap' of system resources. In theory, end-users should not notice any interruption.