An expert car mechanic can get a good sense of the health of your car during a test drive, by observing key performance indicators like engine noise, acceleration lag and vibrations. But to identify the specific cause of poor performance, he needs to look under the hood.
Most modern IT departments managing private cloud or virtualized datacenters use monitoring tools to measure the KPIs of their server infrastructure. These tools typically connect via infrastructure vendor APIs to collect metrics and then display them on a central dashboard.
While these tools provide the big picture of your overall datacenter health and performance, they fall short of looking “under the hood”. These tools are difficult to use for root-cause analysis of performance issues because the APIs restrict data collection intervals and are unable to render granular data.
To really get all the details of how your cloud infrastructure is performing, you need to deploy a very light-weight agent on your compute infrastructure. Agent-based monitoring helps you collect granular data with very short collection intervals, which can’t be rendered by remote monitoring. Such real-time data with high granularity can alert you about possible performance issues in advance and presents experts with the details they need to diagnose and triage performance issues promptly.
For high availability and high performance the recommended approach is to integrate “big picture” datacenter health monitoring with “under the hood” agent-based performance monitoring for drill-down diagnosis and triaging.
How HP Compute Sensor works
HP Compute Sensor is a light-weight cloud instrumentation that provides deep insight into your cloud systems in real time at a 1-second collection interval.
HP Compute sensor is integrated with HP vPV to provide a single user interface for visualizing the overall health of IT Datacenter while also allowing administrators to perform detailed real-time guest OS drilldowns on each virtual machine.
Similar to the Windows task manager running on your PC, the HP Compute Sensor can provide very granular information about the system resources consumed by each hosted application and its associated process. Such detailed insight can only be obtained by an agent-based monitoring solution.
5 ways HP Compute Sensor drills into cloud systems
Although it is less than 5 MB and installs in just 5 seconds, HP Compute Sensor delivers great value for private cloud systems. Its dashboard (Fig. 1) displays:
1. Per-second real-time data collection and streaming
2. Resource-level drilldown views across process, CPU, disk, file system, network, application and system events
3. Process cloud for a quick view of top resource-hungry application processes
4. Process drilldown with deep, granular metrics
5. vPV integration so you can also visualize the big picture of your datacenter with vPV heat maps