Basic component allocation

This example shows two component graph, where allocation ensures that the first component will be executed on cluster node1 and the second component will be executed on cluster node2.

Basic component allocation with remote data transfer

Two components connected with an edge can have different allocation. The first is executed on node1 and the second is executed on node2. Cluster environment automatically ensures remote data records transfer.

Multiple execution

Graph with multiple node allocation is executed in parallel. In this example both components have same allocation, so three identical transformation will be executed on cluster node1, node2 and node3.

Cluster data partitioning

Graph with two allocations. First component has a single node allocation, which is not specified and is automatically derived to ensure minimal number of remote edges. The ParallelPartition component distribute records for further data processing on cluster node1, node2 and node3.

Cluster data gathering

Graph with two allocations. Resulted data records of parallel data processing in the first component are collected in ClusterGather component and passed to cluster node4 for further single node processing.