Proactive monitoring and detail analysis of an Enterprise Hadoop cluster plays a very critical role in successful Hadoop Data Lakes. Monitoring teams, Architects, Devops, Administrators uses a set of monitoring tools and dashboards to ensure that analytical workloads executes as per expected SLAs and Hadoop as a platform is utilized optimally by them. Jumbune offers a Cluster Analysis feature which suites your proactive monitoring, analysis requirements, gives differentiated & niche insights about Applications, Containers, and Cluster. Cluster Analysis feature offers you to perform,
- Proactive Monitoring of the cluster,
- Job level resource consumption tracking,
- Insights of usage patterns of Yarn containers, queues,
- FS major metrics and load distribution across nodes,
- Job wise historical resource consumption analysis,
- User wise Resource utilization analysis on Yarn Queues,
- Real time projection of available containers,
- System and Daemon level JMX metrics and more
Moreover, with Cluster Analysis one can,
Establish Charge back Model - Hadoop as a service platform is one of the prominent use case of creating Hadoop Based Data lake. Enterprise team exposes this platform to different Business Unit, Users and Groups to bring in their use cases of this platform. In such scenarios, the enterprise team may wish to establish a charge back metering model of the utilization of the exposed platform by various use case users. Jumbune helps you to publish a charge back model by analyzing and publishing platform utilization. Reports like High Resource Consumption Business Users & applications, Long Running Applications by Business Users & Applications, Queue utilization behaviors by Business Users are very helpful to derive these models.
Get Real time Recommendations - Jumbune analyzes resource usage patterns of applications, exposed yarn containers and queue utilization to understand usage behavior of the Hadoop cluster, containers, queue utilization and application. With the usage analysis Jumbune produce diverse set of recommendations in real time like,
- Optimal Analytical Workload Schedules,
- Resource utilization for analytical jobs,
- Optimal Configurations for cluster, daemons and queues,
- Job level recommendations for Analytical workloads, etc.
Custom Escalation of Alerts - Alerts on Cluster Nodes, Daemons, File System, Yarn Containers, Jobs can be customized for configurable multiple level escalations to diverse set of recipients on Mail, SNMP and Ticketing Systems.