|
CA’s vision is to unify and simplify the management of enterprise-wide IT. The rapidly growing use of virtualization and clustering in enterprise infrastructures provides a rich environment for increasing complexity. Unlike physically bounded infrastructures, the logical entities in virtualized and clustered environments are not visually apparent.
Nevertheless, you need to know what resources you have and if you are using those resources to their fullest capacity. In your server consolidation initiatives you need to ensure you are making efficient use of resources. In your virtualized environments you need to ensure critical applications receive the right allocation of shared resources. In your clustered environments you need to simplify the management of these critical multi vendor assets. In delivering on Service Level Agreements you need to ensure the services you provide and support are not interrupted and that they meet internal and external customer expectations. In the realm of virtualization this means optimizing the existing IT server investments so that they are service driven, integrated, modular and open. These needs call for a comprehensive management solution.
This paper will briefly discuss the technical management challenges inherent in these complex infrastructures including centralizing management of virtual, physical and clustered environments, virtual sprawl, honoring service level agreements, resource allocation and the prioritization of resources. Then it will demonstrate how CA Virtual Platform Management addresses the specific requirements of these environments, making it possible for you to view and manage logical entities as easily and as well as physical entities. Management Challenges
in Virtual Environments
Heterogeneity adds a great deal of complexity to IT infrastructures and IT infrastructures have diversity on many counts—vendors, software systems, hardware platforms, operating system virtualization, server virtualization and clustering. As time goes on, vendors will likely consolidate and standards will form, causing some organizations to shift purchasing decisions. Choosing a management platform that can support this diversity and flexibility is important.
Virtualization and clustering create unique management challenges, particularly when combined. Hardware no longer defines a logical entity, since a single server can host many virtual machines. Mapping virtual entities—whether multiple, physical machines are operating as one machine or multiple, virtual machines are operating independently on a single physical machine—to physical entities is a must. Not only must they be mapped but they must also be depicted in some visually intuitive fashion. Without this capability you can easily become a victim of the “law of unintended consequences.” While server consolidation may have driven the initial use of virtualized server platforms, in practice there may be some unexpected results. For example, consider an organization that wants to reduce their server deployment from 300 to 30, where each server hosts 10 virtual machines. Virtualization makes it easy to configure additional virtual machines. The result can be that virtual machines begin to proliferate, resulting in what is often referred to as “virtual sprawl”. As a result, IT organizations find it difficult, if not impossible, to keep track of all of these virtual machines as they continue to propagate, so that 300 virtual machines on 30 servers turn into 450 virtual machines on 30 servers.
Operating System Virtualization
Virtualized operating systems run on a server with a single host operating system. Multiple operating systems are not supported but overhead is reduced.
Server Virtualization
Virtualized hardware runs multiple guest operating systems. The server does not have a host operating system. Multiple types of operating systems are allowed but overhead is increased.
Clustering
Several locally-attached physical machines provide distributed processing power, while appearing as a single processing resource. It becomes more complicated when:
- multiple vendor technologies are used
- clustering is used within a virtualized environment
Resource optimization is one of the primary reasons people decide to incorporate virtualization, but resources are not automatically optimized without management. In fact, with the ease of creating virtual machines already discussed and without a carefully considered management solution, resources can become scattered. Inactive virtual machines “disappear” taking their resources with them.
Allocated resources must be “retrieved” so they can be reallocated, if the promise of virtualization is to be achieved. Enough resources must be conserved so the hypervisor, the software that makes virtualization possible, and other essential processes have the CPU and memory resources required to run. While trying to avoid over-utilization, organizations can fall back into underutilization— just the problem they were trying to correct.
|
