High Performance Computing Using In-Memory Massively Parallel Processing
The GridGain® in-memory computing platform includes a distributed compute grid which supports high performance computing. The GridGain compute grid distributes CPU or otherwise resource intensive processing tasks across the underlying grid of server nodes. This provides support for high performance computing (HPC) and massively parallel processing (MPP) use cases.
Queries are divided into sub-queries that are sent to the relevant nodes in the compute grid of the GridGain cluster. The sub-queries are then processed on the local CPUs where the data resides and sub-query results are aggregated to produce the results for the query. Parallel processing the sub-queries results in much faster performance than is possible by running the query on a single node.
GridGain is easily scaled out by adding server nodes to the cluster. GridGain can also be scaled up by upgrading the cluster with more powerful servers. Scaling up in addition to scaling out the GridGain cluster allows users to create extremely powerful capabilities for high performance computing built on an in-memory computing platform. Ultra low latency use cases can be addressed using high performance Java Virtual Machines (JVMs).