Among the many pieces you have to play on the IT cost-savings chessboard, virtualization with PowerVM could be the rook. An incredibly powerful piece, putting the Rook into play can yield many benefits and shore up an IT department's efficiency in a defensive effort much like that of castling. In this article, I want to spend a little bit of time talking about why you should consider virtualizing your IBM i on Power Systems environment. For me, the simple answer to why you should leverage virtualization is the ability it gives you to consolidate workloads, maximize resource utilization, build a reactive environment, and streamline workload management.

You can find a technical look at the virtualization capabilities of IBM Power Systems, as well as the virtual I/O capabilities of IBM i and VIOS, in the article "IBM Systems Have the POWER of Virtualization Built In: A Technical Look" (ID 63186).

Workload Consolidation

Most enterprises have a mix of operating system deployments—from Windows and Linux to IBM i and AIX. When these operating systems and their related workloads are deployed on divergent hardware, the IT staff has to deal with the inevitable issues of different backup mechanisms and hardware resources (such as memory and processor) that cannot be shared (or moved) among systems, as well as space and power limitations. Consolidating those workloads onto a converged hardware platform minimizes all those concerns. The backup mechanism can be streamlined through the sharing of a physical backup device, as well as through the ability to treat virtual I/O as an object that can be backed up for disaster recovery purposes. Memory and processor can be right-sized for the workloads and moved between workloads either at the discretion of the IT staff or automatically through the system firmware. And consolidating workloads on a reduced number of physical footprints reduces space and power concerns.

Each of the supported operating systems has its own purposes within an enterprise. Typically, companies use IBM i to deploy line-of-business applications such as ERP packages. Those that employ Linux often use it to deploy infrastructure solutions such as file serving, web serving, and firewalls. Shops running Windows tend to tag it as the network operating system (NOS) for the network infrastructure. Windows also is used to provide complementary or front-end applications for the IBM i back-office applications and to deploy some line-of-business applications. Finally, IBM i shops that deploy AIX tend to use it for one of three types of applications—Oracle database and/or application server, WebSphere, or AIX-based ERP/SCM applications.

Maximize Resource Utilization

Deploying multiple workloads on a converged hardware platform lets you maximize a system's resource utilization by right-sizing the workload. Keep in mind that each workload in a logical partition (i.e., IBM i, AIX, and Linux) makes use of the same processors and memory on the Power System. This helps you avoid the issue of non-compliant resources between systems and lets you specify the portion of the overall resource that each workload requires. For example, if—through sizing—you determine that an IBM i partition requires 4 ½ GB of Memory and 1 ¼ processors, you can allocate just that amount of resource to the partition. Then, if the system actually has 16 GB of memory and 2 processors, the remaining memory and processor would be available to other workloads. Additionally, you can define hardware resources such as tape controllers (and their associated tape devices) as shared devices and logically move them between workloads. There is no need for physical movement and recabeling in this environment!

Build a Reactive Environment

Consolidating workloads on a converged hardware platform also lets you establish an environment in which you can react to resource requirement changes as they occur. For example, if one workload experiences a spike in resource requirement, the server can move the necessary resource from another workload that doesn't require it. Additionally, you can configure the logical partitions so that the hypervisor monitors processor utilization requirements and makes additional processor resources available as needed.

And when the logical partitions are properly configured, resource allocation changes can be made while the workloads are active.

Streamline Workload Management

Hosting all these divergent workloads on a converged hardware platform provides the foundation to streamline management of the entire infrastructure as well. For example, for all of the workloads that leverage virtual I/O, the method for creating the "disk" that the operating system will use—as well as for backing up that disk for disaster recovery—is the same. Additionally, when virtualizing I/O via IBM i, the object that represents the disk for the operating system can easily be copied, allowing for creation of additional instances without the need to reinstall the operating systems. For IBM i, Linux, and AIX workloads, logical partitions are created to define both the hardware resources as well as the virtual adapters and relationships to support virtual Ethernet and Virtual I/O. The creation of the logical partitions for each of these workloads is performed through the same management tool (see "IBM Systems Have the POWER of Virtualization Built In: A Technical Look" in the ProVIP section of this magazine (page ProVIP XX, ID 63186) and is the same function regardless of the workload. You can also use tools such as IBM Systems Director to manage operating system events and resources for each of the deployed workloads.

The HA Concern

Entering into a virtualized world is not without its concerns, and proper planning should be undertaken. When resources are hosted on a single hardware platform, it introduces a possible single point of failure into the environment. However, you can mitigate this concern through proper planning and implementation. For example, when you leverage virtual networking to support multiple workloads, the virtual IP addresses (VIPA) capability of the IBM i operating system could be used to route the virtual network traffic over multiple physical network adapters. This way, if a physical adapter fails, another physical adapter can take over the routing of the traffic between the virtual and physical networks.

Make a PowerVM Play

Put your most powerful piece (Power Systems virtualization) on the IT cost-savings chessboard into play, and reap the economic efficiencies of workload consolidation, resource utilization, a reactive environment, and streamlined workload management.

Erwin Earley is a managing consultant at IBM who has worked with the Rochester, Minnesota, development lab since 1996. Erwin currently heads up the Open Community Center of Competency in the IBM i Technology Center. He has worked in the IT industry since 1980 and has experience with several Unix variants as well as Linux and IBM i.

Editor's Note: Check out the technical companion of this article, titled "IBM Systems Have the POWER of Virtualization Built In: A Technical Look" (ID 63186).