The planned purchase of Platform Computing by Big Blue (reported on CloudPro this morning) raises a few questions on how IBM will continue to operate in the HPC and cloud market. On the one hand IBM has invested heavily in xCAT, the cluster management system, which powers some of the largest supercomputers in the World such as RoadRunner, to name one. However, its planned acquisition of Platform Computing (which has also spent a lot of time in the last year to improve its product) throws that investment into question.
Through Platform Computing, IBM will now have a cluster manager, a full-fledged scheduling system (using Platform Computing’s Load Scheduling Facility), and a front facing portal with monitoring and statistics. This acquisition also provides IBM with an MPI suite (Platform MPI) which is used with many applications across commercial and research facilities.
The past investment in xCAT and new investment in Platform Computing begs the question: how is IBM going to combine these two investments? Will IBM just use Platform Computing’s cluster manager as it is? Will it remove the cluster manager and replace it with xCAT?
In terms of schedulers /resource managers i.e. IBM’s LoadLeveler versus Platform Computing’s LSF, which one will IBM ship? Will it be bringing both together, and use the best bits of both? My personal guess is that IBM will make LSF compatible with Loadleveler, and xCAT, and maybe give customers the choice.