Bright Computing has announced the next generation of Bright Cluster Manager. The system employs Bright roles to make it easier to configure Hadoop services such as HBase, YARN and ZooKeeper.
Bright Cluster Manager collects metrics and performs health checks for Hadoop that target HDFS, JVM, job and network activity. Bright can be used to integrate with MapReduce and YARN to manage Big Data workloads efficiently, and ensures high availability of vital Hadoop services.
Bright has also implemented a revision-control feature for the software images used in provisioning clusters and clouds by using btrfs snapshotting or cloning.
It also employs cloud utilisation features, Bright Cluster Manager works with storage services native to Amazon Web Services (S3 and Glacier). Bright instantiates compute resources in the cloud only when data is actually ready for use by applications. It also automatically starts and stops nodes in the cloud based on workload in the queuing system.
Bright has stated that the Bright cluster manager now has full support for Cisco Unified Computing System (UCS). This means it can manage bare-metal BIOS settings and multi-node firmware upgrades, out-of-band services (e.g., power) and integrate them directly with the Cisco Integrated Management Controller (CIMC) to configure network, identity and other services. Bright Cluster Manager samples metrics and incorporates various health checks specific to Cisco UCS rack Servers.