For quite some time we have taken the position that in order for the next 60% of the workloads and applications to get virtualized, that the staff operating the virtual environment is going to have to take responsibility for the performance and availability of the applications running on that infrastructure. The logic behind this is simple. If you want someone who owns a performance critical application to give up their dedicated hardware and move into a shared service environment, you are going to have to guarantee the performance of their application to them in order to be allowed to virtualize that application.
It is now clear that VMware (at least conceptually) understands and agrees with this point of view. At VMworld Copenhagen VMware made a series of management announcements that both clarified VMware’s long term management strategy, and put some specific product announcements behind that strategy.
There are two overriding themes to VMware’s application management strategy. The first is that monitor->alert->fix manually, gets replaced by monitor->fix automatically->notify the humans, and the second is that VMware intends to manage the entire stack in the virtualization environment ranging from the storage array to the applications. Let’s take a look at what is new.
New Version, Packaging and Pricing for vCenter Operations Manager
In Copenhagen VMware announced a new version of vCenter Operations Management Suite, version 5.0. Â The key new features are:
- An enhanced dashboard that leverages the Integrien technology to a greater degree to provide for Health, Efficiency and Risk metrics about the environment.
- Full integration of CapacityIQ into the Advanced and above editions of the product
- Overlay of in-guest change events from vCenter Configuration Manager
- Integration with Chargeback Manager for Cost-Based Capacity Optimization
- VMware View specific features like PCoIP performance monitoring
- Application discovery and application dependency mapping with vCenter Infrastructure Navigator
New vFabric AppDirector Product
AppDirector constitutes VMware’s foray into applications provisioning and management. This is a significant step beyond the previous focus upon images and vApps. The idea here is to introduce Application Templates that contain the knowledge needed to deploy specific applications into specific environments and then to use these templates to fully automate and manage the application deployment process. This is depicted in the diagram below.
New vFabric Application Performance Manager
In “VMware Joins the APM Revolution with vFabric Application Performance Manager“, we detailed many of the new features of vFabric APM, and most importantly why the APM industry needs to reinvented due to the new requirements imposed by virtualization, IT as a Service, Private Clouds and Public Cloud Computing. The most important part about vFabric APM is that VMware built it from the groud to address these new use cases. A good example of this is the ability to automatically detect new applications, map their topology and instantiate end-to-end response time monitoring. This is depicted in the diagram below.
So What to do and How to Pick?
VMware is certainly well on its way towards becoming a serious vendor of management software for its own vSphere platform. It is in this focus upon just the vSphere platform that the first serious choice arises for the buyers of management solutions. Management in the physical world is a horrible mess because management solutions of all types got purchased multiple times – once for each platform or type of product. If you currently are, or are likely to become a company with more than one virtualization platform you should seriously consider not repeating this mistake when it comes to how to manage your virtualization platform and pick a management stack that works across your virtualization platforms of choice.
So if you are considering alternatives to VMware management solutions how should you proceed. Let’s break this down by category.
At the level of Operations Management, the focus isÂ fundamentallyÂ upon managing the performance, capacity, and configuration of the virtualization environment. Quest Software (vFoglight) and Veeam (Veeam One and nworks) have market leadership positions in terms of the number of installed customers. vKernel offers a compelling and very easy to use performance and capacity management product that supports both vSphere and Hyper-V. VMTurbo offers a compelling free solution, and the paid for solution offers unmatched abilities in the area of workload prioritization, and automated placement of workloads to achieve desired service levels. Netuitive is the only real alternative in the area of self-learningÂ analyticsÂ and offers direct integration with the vSphere API’s and adapters into just about anything else you could possibly own. Reflex Systems has combined very granular management of security and configuration with performance and capacity monitoring that has been proven to scale up into some of the largest vSphere environments on the planet. SolarWinds certainly earned a look by winning the Best of VMworld Las Vegas aware in this category in September. Liquidware Labs and Xangati still offer more features when it comes to managing VDI in production, and offer the front end assessment tools that help you plan VDI as well.
At the level of Applications and Applications Performance Management the choices get even more difficult. The first question you have to address is do you want to automate your application provisioning and if so, do that just for vSphere or for other environments as well. rPath is a good alternative for a solution that fully supports vSphere but also allows you to provision applications to physical machines and various public clouds.
Your next tradeoff will come in the area of breadth vs depth, in Application Performance Management. The questions here are 1) are your applications purchased, custom developed, or both, and 2) to what applications frameworks have your applications (whether purchased or custom developed) been built? This leads to the question of why you are buying an APM solution. If the reason is to find bugs in code in production for a custom developed application then you need an “inside the runtime” solution that is basically an agent inside of your application. This kind of solution is only available for certain kinds of applications like Java, .NET, PHP, and Ruby. If this is not the case and you have all kinds of applications written to all kinds different environments then you will need something more general but you will give up code level visibility in the process.
The diagram above depicts these tradeoffs. It is easy to get code level visibility for a certain set of applications, an you can also get end-to-end response time visibility for broader applications by trading off that code level visibility.
The Role Application Operations
In “Is it Time to Reorganize Data Center Operations?“, we proposed that Data Center Operations needs to become “Virtual Operations” and that the fragmented groups that support applications need to be consolidated into “Application Operations”. This is the group that will need to support these applications in production in order for the next 60% of the applications to get virtualized. It is clear from the conversations with the VMware product team that VMware is fully on board with this concept as well.
VMware is clearly serious about extending its management reach into the application layer of the stack. New products like vFabric AppDirector, and vFabric APM make this into more than intention – they make VMware’s participation in these markets a reality as well. The move to remake the APM industry has thus far been lead by creative start-ups like New Relic, AppDynamics, BlueStripe, dynaTrace and ExtraHop. Now VMware has joined this effort will add considerable mass and velocity to the effort. The message is clear – it is time to instrument all of your applications for response time and the legacy APM vendors do not have the products that are up to the task.