VMware has announced that it has delivered vCenter Operations Suite 5.0. This is a significant step forward for the management of virtualized environments because it consolidates many of the functions traditionally into a suite that is least somewhat integrated (and becoming more integrated over time). The key thing that VMware is bringing to the table is the integration of resource based performance management resource based capacity management, configuration management, and a brand new vCenter Infrastructure Navigator (VIN).
The New vCenter Operations Console
One of the very nice new features of vC Ops 5.0 is a reworked and dramatically easier to understand top level console. There are now top level scores for Health (the current state of the environment), Risk (issues caused in the future primarily by running out of capacity), and Efficiency (cases where things have been over-provisioned and you are wasting money). All of these scores leverage the metrics that VMware collects from its own virtualization platform and the analysis of these metrics by the Integrien technology acquired in 2010.
With this release, the former CapacityIQ product is fully integrated into vCenter Operations Advanced, Enterprise, and Enterprise Plus (more on the editions later). The next piece of integration is the still partial integration of vCenter Configuration Manager into vC Ops Enterprise and Enterprise Plus. We all know that a very high percentage of the performance and reliability problems in an environment (especially a dynamic one) is related to changes in configuration. In the example below, the configuration change tracking feature in vC Ops 5.0 found an issue with storage in the Hands on Labs at VMworld 2011.
One of the core issues with all infrastructure management solutions is the degree to which they either have or do not have visibility into the actual applications, applications systems, and business services running on the infrastructure. First generation management solutions on the physical side completely lack the ability to understand application inventory, application topology, and application dependencies on a real time basis (necessary to keep up with a dynamic environment – especially a private cloud).
Infrastructure Navigator is therefore a huge addition to the Enterprise and Enterprise Plus editions of vCenter Operations 5.0. Infrastructure Navigator not only discovers your applications, it shows you how pieces of them communicate with each other, and where those pieces are running right now. The ability to do all of these things automatically and in real time should for once and for all put the nail in the coffin of the CMDB’s and the legacy management frameworks that rely upon them.
The Editions of vCenter Operations
vCenter Operations comes in four editions:
- Standard Edition includes just the resource utilization based understanding of current performance. It is intended for relatively small and not very complex environments. Standard supports only vSphere as a virtualization platform.
- Advanced Edition adds what was formerly CapacityIQ – the resource constraint based understanding of current and future capacity. Advanced supports only vSphere as a virtualization platform.
- Enterprise Edition adds Infrastructure Navigator and Chargeback and is a new product offering. It basically contains all of the features of the old Enterprise (now renamed Enterprise Plus) but those features and the support of the product are limited to just the vSphere platform.
- Enterprise Plus is the old enterprise and is the version that with the adapters into third party products and lots of professional services can support other virtualization platforms and even non-virtualized servers.
Issues and Alternatives to Consider
While vCenter Operations 5.0 plows considerable new ground in terms of a management solution for vSphere there are issues and limitations which should cause you to consider third party solutions in many cases:
- Pricing. While we did not get into price in this article, the per VM pricing for vCenter Operations is an issue for many customers. The bottom line is that if you have 10 VM’s per server and deploy Enterprise Edition you will pay around $195 per VM or $1,995 per host. When you take advantage of the efficiency gains delivered by the product and double your density to 20 VM’s per host you will pay an additional $1,995 to realize that gain. None of VMware’s competitors on the resource and availability management front like Quest Software, Veeam, ManageEngine, Netuitive, Phd Virtual, Reflex Systems, Solarwinds, VMTurbo, Xangati, or Zenoss price this way and therefore none of them implement a “Density Tax” in their products.
- Scalability issues. One of the things that is not known at this point about vCenter Operations is how well it will scale up into very large environments that have hundreds, thousands, tens of thousands, or even over one hundred thousand VM’s. Management at this level is a completely different beast. Certain vendors like Netuitive and Reflex Systems have invested years in building highly scalable solutions for the most demanding use cases, and we just do not know how far VMware is up this curve yet.
- Where is the Automation? At VMworld 2011 Paul Maritz laid out a vision of automated IT Operations driving reductions in OPEX through “monitor, fix, notify the humans”. This is an extremely ambitious strategy and an well integrated platform like what vC OPS is going to be after Configuration Manager is fully integrated is an essential starting point. But the fact is that right now, VMTurbo is the only vendor that can automatically adjust how the resources in the environment are allocated to workloads to ensure workload performance on the basis of priorities.
- Integration. We mentioned at Configuration Manager is not fully integrated yet. Therefore Enterprise and Enterprise Plus are still at the moment more bundles than they are fully integrated suites. This will certainly get addressed by VMware over probably the coming year – but it remains a consideration right now.
- Platform Support. It is increasingly clear that many organization either have or are going to end up with multiple virtualization platforms. One of the reasons that management is such a mess in the physical world is that people bought a management product for every infrastructure element and application type that they owned. If you are going to have more than one virtualization platform you should absolutely not build two management stacks on top of those platforms. This is where a vendor like Microsoft with the nworks plugin from Veeam and the APM plugin from BlueStripe has a tremendous advantage.
- Deeper APM and Infrastructure performance visibility. The bottom line is that discovering applications and measuring their performance via looking at resource utilization metrics is a legacy old school way of doing things that stands no chance of working for business and performance critical applications in virtualized and cloud based environments. The performance of an infrastructure can only be understood by measuring its end to end latency. This is where vendors like Xangati and Virtual Instruments shine. On the applications front a new set of APM vendors like AppDynamics, BlueStripe, dynaTrace (now part of Compuware), ExtraHop Networks, and New Relic have brought to market solutions that are much easier to deploy and much cheaper to own than first generation legacy APM solutions. VMware has announced vFabric APM, its own APM offering, but much work remains to be done on this front by VMware.
- Real Time, Continuous and Comprehensive visibility. The bottom line is that vC Ops is reliant upon the rollup interval for your data in vCenter. That is usually either 20 minutes or 5 minutes. Once your infrastructure grows to the point where a product like vC Ops Enterprise is justified, way too much can go wrong in 4 minutes and 59 seconds. One of the things that invalidates the legacy management frameworks is their reliance upon polling interfaces on 5 minute like intervals for data. The correct approach is to focus upon real time (right now), continuous (all of the time) and comprehensive (do not miss any spikes or anomalies due to sampling) data collection approaches. This is where the modern APM vendors like AppDynamics, BlueStripe, dynaTrace (now part of Compuware), ExtraHop Networks, and New Relic, shine at the applications level, and where infrastructure management vendors like Reflex Systems, Xangati and Virtual Instruments shine at the infrastructure layer.
The delivery of vCenter Operations 5.0 to the market by VMware represents several important milestones in Operations Management for virtualized environments. The tight integration with vSphere and the integration of the Integrien analytics with real time configuration change detection and application mapping put a nail in the coffin of legacy solutions that rely upon periodically updated CMDB’s to understand the environment. The integration of performance, capacity, configuration, and application discovery first into a bundle and later into a full suite will put pressure on many vendors of point solution. However the biggest question remains the viability of a management strategy focused just upon vSphere, when every other vendor (including Microsoft) is taking a more open and cross-platform approach.