Rack-aware scheduling

Using user-defined scripts that map the network topology of your cluster (including HDFS), IBM® Spectrum Symphony intelligently allocates hosts based on the locality of input data for map tasks and schedules waiting map tasks on idle hosts within the same rack or with the best affinity (for example, those with the shortest network distance).

Use rack-aware scheduling especially for input-heavy workload, where network issues affect allocation of hosts with the input data for map tasks. In this case, if hosts with the input data cannot be allocated, hosts closest to the input data are allocated.