IBM infrastructure solution for High Performance Computing

Daniele Piccarozzi



There is an evolution occurring in the demand for HPC. In addition to classical algorithm based computing, sophisticated analytics on Big Data is crossing over into the domain of HPC. Systems need to be able to adapt to effectively address both. High Performance Data Analytics is beginning to require the same magnitude of compute resources, so it is vitally important to have a system which has compute and storage components that can address both. Storage systems also need to be of such design that the same media can support different access methods in a reliable, yet flexible manner. A common storage solution which can support parallel I/O access methods, needs to also very affectively handle unstructured data access from platforms such as Hadoop. Data volumes are growing so fast that the storage system has to be scalable from a capacity and performance perspective. In addition, there are analytics that require large in-memory data needs.

There are innovative ways to use technologies such as Flash to accomplish this. IBM is driving this transformation through development and implementation of leading edge technologies, including hybrid accelerated architectures, innovative storage solution and hpc cloud computing services.