When VMware announced the three editions of vCenter Operations, VMware sent a couple of very clear messages about how VMware felt that monitoring solutions for vSphere should be constructed. The first message was that VMware views Performance Management and Capacity Management as two sides of the same coin. The second message was that Configuration Management as an essential part of a performance and capacity management solution since so many of the problems are in fact configuration related. The last message was the given the complexity and rate of change in virtualized environments that the interpretation of monitoring data has to be automated with self-learning analytics.
VMware vCenter Operations Enterprise embodies all three of these concepts. Performance and Capacity Management are integrated via the inclusion of CapacityIQ. Configuration Management is addressed via the inclusion of vCenter Configuration Manager. Both CapacityIQ and Configuration Manager are today only bundled in, not truly integrated, but this will change over time. And of course, the interpretation of the monitoring data is automated via the inclusion of the self-learning analytics from Integrien, which automatically produce Workload, Capacity, and Health scores. The version of the Integrien technology included in vC Ops Enterprise also supports the existing adapters to third party solutions, and VMware has stated that they will build custom adapters for enterprise customers on a professional services basis if required. So the bottom line message of vCenter Operations Enterprise is, “here is an integrated approach to performance, capacity and configuration management that features self-learning analytics that are open to whatever other monitoring data you want to include in the system”. This is exactly the right message for a vendor that is serious about the enterprise performance management space in virtualization to be delivering.
In Virtualization Performance and Availability Monitoring – A Reference Architecture, we proposed that a horizontally layered approach be taken to monitoring the hardware and software stack in the virtualized environment, with potentially separate products for the storage, LAN, server, virtualization platform and applications layers, with the data feeds from the products at each of the layers integrated into self-learning technology. If we update that post with the concepts that performance, capacity, configuration and root cause need to be integrated with such a layered approach we get the diagram below.
The VMware Answer
Courtesy of the vCenter Operations announcement we know exactly what VMware’s answer is to the requirements posed by the diagram above. Buy vC Operations Enterprise, and to the extent to which it does not have the data that you want in the self-learning model, buy an adapter, or pay VMware to build on for you that feeds your desired data into the vC OPS self-learning model. This would look like the diagram below.
Third Party Alternatives
But what if you wanted to pursue the approach described by the diagram above, but do so in a way that supported the VMware vSphere platform, but was independent of the platform (and independent of VMware as a management vendor). Could this be done, and if so how. The answer is yes, and the remainder of this article is about how. If we were to populate the boxes in the diagram above with the vendors who fulfill this roles, it would look like this.
The point is that there is no shortage of third party solutions that can be assembled into a picture like the one above and integrated via a self-learning system like Netuitive. For that matter, there is nothing conceptually wrong with assuming that any of the vendors in the diagram above who have monitoring data of interest to you could have their data feeds integrated into vC OPS Enterprise either. The important point is that you need to acquire products that comprehensively handle their layer of the stack, and then integrate with with the self-learning technology of your choice.
The questions of which vendors are best suited to be chosen at each layer and why will be dealt with in subsequent posts. Note that this article focused just on alternatives to vC OPS Enterprise, not vC OPS Standard or Advanced. The question of how a mid-sized account with 500 VM’s or less should approach performance and capacity management will also be dealt with in separate articles.