Multiple scheduling schemes
A fair-share scheduling scheme with 10,000 priority levels can be used for multiple jobs for a single application. Preemptive and resource-threshold-based scheduling is available with runtime change management. Multi-dimensional scheduling is supported with the ability to specify slot definition for each consumer in terms of up to four customizable resource metrics.
The resource-sharing model in Spectrum Symphony helps reduce infrastructure expenses and management costs. It enables you to deploy multiple heterogeneous applications on the same shared grid—and at the same time, preserve line of business (LOB) ownership while delivering service-level guarantees.
Built in Hadoop MapReduce-compatible implementation
The Spectrum Symphony Advanced Edition includes an Apache Hadoop-compatible MapReduce implementation optimized for low latency, high reliability and resource sharing. Using this capability, you can run Hadoop and non-Hadoop applications on the same shared distributed infrastructure. The multi-tenant architecture allows multiple MapReduce engines to be deployed on a single shared infrastructure.
Supports multiple clusters
Multiple cluster support allows a set of resources to span multiple clusters and to be moved between clusters based on need. A single session can run across multiple clusters, both on-premises and on-cloud, efficiently managed via a single interface.
Supports rolling upgrades
With rolling upgrades, administrators can easily install and configure a newer version of Spectrum Symphony over an older version without the need for manual migration.
Three editions of IBM Spectrum Symphony
Spectrum Symphony is available in three editions, all featuring low-latency, high-performance computing service-oriented architecture, agile service and task scheduling. The editions range in scalability from one or two hosts for Developer Edition, to up to 5,000 hosts and 128,000 cores for Advanced Edition—your best choice for distributed compute- and data- intensive applications, including Hadoop MapReduce. The Standard Edition offers enterprise-class performance and scalability.