Geode on Virtualized Infrastructure: How to Succeed with Geode on vSphere & Public Clouds
Recently there has been a surge of interest in running Geode workloads on virtualized environments. In most cases, operations teams attempt to first manage Geode workloads the same way that they manage application server workloads, which almost always leads to cluster stability and service availability problems. Tactical attempts to improve Geode workload stability via vSphere resource reservations only address part of the problem and rarely produce satisfactory results. The right way to deploy Geode onto vSphere infrastructure is by first carving-out dedicated Hardware Resource Pools, isolated from Application Server workloads and with more appropriate target utilization rates and consolidation ratios. In this presentation we'll discuss the fundamental differences between Geode and Application Server workloads as they relate to virtualized infrastructure management, focusing on differences in defining and measuring target SLO's, capacity planning, and on how to communicate these differences to vSphere administrators. We'll also describe how to diagnose common virtualization-related performance and stability problems from application behavior and performance metrics.
Principal Data Engineer, Pivotal