IBM NeXtScale System, Планирование и руководство по внедрению

Страница: 237 / 266

Chapter 8. Software stack
219
򐂰 Uses xCat technology and offers greater cluster management scalability by
scaling up to 1,000 nodes in a cluster via a centralized, web-based user
interface.
With Platform Cluster Manager Standard Edition, administrators can define
provisioning templates on cluster nodes. Adding or removing software
components can be done without rebooting the operating system.
IBM Platform Cluster Manager Advanced Edition
IBM Platform Cluster Manager Advanced Edition deploys and manages multiple
analytics and HPC clusters to deliver the environments and performance that are
needed by the most demanding compute-intensive and big data workloads. The
Advanced Edition includes the following main features:
򐂰 Automated provisioning of physical and virtual clusters.
򐂰
Consolidation of separate HPC and analytics clusters onto a shared
infrastructure.
򐂰 Policy-based cluster scaling, growing, and shrinking a running cluster as a
foundation for building a dynamic shared analytics and HPC cloud.
򐂰
A management portal that enhances administrative and user productivity,
which expands the ability to meet and exceed service levels and lower
operating costs.
򐂰 Workload monitoring, resource monitoring, alerting, and troubleshooting.
򐂰
Integration with the xCAT (or multiple instances of xCAT), with which
hundreds of servers to be managed by one administrator.
The integration of Platform Cluster Manager Advanced Edition and xCAT
delivers a highly scalable and automated resource management capability. The
management portal increases the ease of use for multiple xCAT instances,
where server inventories can be viewed and monitored from a single pane of
glass. The policies and cluster definitions of Platform Cluster Manager Advanced
Edition automate the provisioning capabilities of xCAT and together they create a
flexible environment to rapidly deploy, grow, and shrink clusters.
Through a single self-service interaction, users can deploy entire running
analytics and HPC clusters that are ready to use in minutes, instead of days or
weeks, including the master, compute and support nodes, and portal
applications, MPI drivers, and the application software. Administrators can use
the portal to rapidly deploy a mix of technologies, such as, IBM Platform Load
Sharing Facility (LSF), IBM Platform Symphony, Hadoop, and most other
third-party workload managers.