|
As you start moving to large scale virtual environments with hundreds or thousands of virtual servers, you need to manage it well. Management of a virtual environment presents unique challenges; many organizations do not thoroughly anticipate the differences in managing a virtual environment from a purely physical environment. Many of your existing management tools may, in fact, only be able to manage physical machines or may collapse under the sudden load of many more systems.
In virtual environments, discovery, visualization and optimization are done differently than they are done in purely physical environments. For example, discovery of physical machines or hosts is a single step process. In addition, any discovered machine can be traced back to a paper trail that shows the server purchase and the footprint it currently occupies in a data center. In a virtualized environment you have to discover the machines within the machines, and keep track of them.
The complexity of today’s IT infrastructures and the dependence of businesses upon the performance and availability of IT assets require that your IT staff understands the relationships between business services, physical hosts and virtual machines. Visualization of the relationships between these components must be accurate and must extend end-to-end. Adding to this challenge is managing clusters of virtual machines that reside within the host. To optimize your resources you must track the performance and utilization of the host machine and the virtual machines (VMs). Optimizing within a host and optimizing across the infrastructure are at the core of maximizing the value of your virtual environment.
Discovery and Visualization Challenges
If you do not know what you have you cannot manage it. Discovery, monitoring and management of virtual environments are critical functions and you need the right tools in place to perform them. Ideally, virtual platform management should include advanced capabilities with respect to discovery and visualization. It should automatically discover virtual and clustered resources and create intuitive visual maps, which relate the virtual elements to physical elements, as well as related business processes.
Traditional discoveries are topology aligned and conduct subnet sweeps. The dynamic nature of virtualized environments necessitates a different discovery mechanism. For example, a host may be in a different subnet than its associated sessions, so when you visualize them it is important to have their relationships understood. When migrating sessions, it is easy to lose track of them; the right tool will maintain the appropriate relationships in real time. Another aspect of visualization is being able to see and report on the performance of the guest OS and the physical OS. You need the ability to look at both levels and view capacity from both perspectives.
Optimization Challenges
Resource optimization is one of the primary reasons people decide to incorporate virtualization, but virtualization does not automatically optimize resources. In fact, the opposite can be true. Without a carefully considered management solution, resources can become scattered. Inactive virtual machines “disappear” taking their resources with them. Previously allocated resources must be “retrieved” so they can be reallocated, if the promise of virtualization is to be achieved.
It is ironic that a compelling reason many companies employ virtualization is to reduce server sprawl and get better utilization out of their server resources, yet the ease with which organizations can build, duplicate and deploy VMs exposes the organization to the proliferation of VMs — virtual sprawl — so you must be alert to the Law of Unintended Consequences that replaces physical server proliferation with virtual server proliferation. Through continuous discovery mechanisms, visual maps can be constantly kept up-to-date — preventing the “lost server” problem, and mitigating the impact of server sprawl.
Another consideration in resource optimization is resource conservation for essential processes, including the hypervisor, so they have the CPU and memory resources required to run. If organizations don’t have the advanced discovery, visualization and management tools to be confident of their resource allocation, their resulting caution can push them back into under-utilization — just the problem they were trying to correct.
Resource allocation and reallocation needs to happen all the time and it is essentially impossible to manage as a manual task. Policy-based automation makes resource optimization a dynamic function, allocating and retrieving resources and maximizing use of the resource pool.
|
