Why is Linux the answer for data center consolidation? To start, for a large IT infrastructure project to be approved in today's climate, it must make real business sense. The project must have a significant return on investment (ROI), which will lower total cost of ownership (TCO). Linux operating systems work well with these demands, as will be explained in this tip.
The primary goal of any data center consolidation project is to lower power and cooling costs, reduce real estate footprint and improve overall server workload utilization. Because data center consolidation usually involves server consolidation, this type of project should also substantially reduce the cost of service contracts and software licensing, while substantially increasing overall server workload utilization. I've led projects in which the largest single cost reduction was on the software licensing side. (Those of you that have worked with Oracle licensing on your server farms know what I'm talking about.)
Data center consolidation strategies
There are essentially two ways to consolidate data centers. One is a like-for-like forklift move. For example, a customer looking to consolidate four data centers into two. This scenario simply involves moving all of the equipment from the four sites into two data centers. There may be some virtualization and server consolidation involved, but no real effort to redesign or reengineer infrastructure. The other scenario would be to increase efficiencies and processes along with the consolidation of data centers. This involves strategic planning regarding the overall direction of infrastructure. While more complex and difficult, the latter move is the one that will provide the greatest ROI and TCO.
 |
|
|
|
|
Linux runs on more types of hardware platforms than any other operating system.
|
|
|
|
|
|
|
|
|
|
So why should Linux be part of a data center consolidation plan? To start, it's important to note that Linux runs on more types of hardware platforms than any other operating system. This include everything from x86 commodity hardware, to Unix RISC based systems from HP, Sun and IBM to the IBM System z (mainframes). In a typical enterprise environment, there may be up to half-a-dozen operating systems to support. This might include Windows, Linux, several flavors of Unix and zOS (on the mainframe). As you know every OS requires staff to support that system: operators, administrators, engineers, capacity planners and architects. A typical enterprise organization may have at least a dozen folks supporting each OS. If you're supporting six operating systems, that's a staff of 72 people.
Operating system consolidation
If your data center consolidation management strategy is built around Linux, you can keep your mainframe but run Linux. Same thing with your IBM p595 Power servers (formally running Unix) and your commodity x86 blades. At the same time you may also consider migrating off of Solaris to Linux. These moves can streamline your infrastructure. While you will not likely be able to trim staff down to just six people, but you might be able to get by with half of the staff previously needed. This is just one small example of how you can increase efficiencies by doing an operating system consolidation project in conjunction with a data center migration. As you know, Linux is no longer something just resigned to the back-end of data centers running Web or DNS servers. In fact, many companies already run all their information systems on this platform (e.g., Oracle).
In regard to the mainframe, running Linux on the mainframe has become increasingly popular in recent years. This is because in many ways it gives customers the best of all worlds: the dependability of 40 + years of hardware maturity and resiliency, along with a flexible resilient open source operating system. Many customers actually use Linux on the Mainframe as the basis for their entire data center consolidation strategy. We've come full circle from the days of decentralization with a server under users desks, all the way back to the data center.
For customers that don't have Mainframe expertise, you can do a similar effort using High end RISC servers that formally ran only Unix. IBM's POWER servers are one example, and you can run either IBM's Unix, AIX, or Linux (SUSE or Red Hat) on their System p platform. Both operating systems support IBM's PowerVM, which provides many elements that lend themselves to consolidation of systems. This includes the ability to use Virtual I/O servers, shared processors and take advantage of recent innovations such as Live Partition Mobility, a technology that allows movement of a running partition from one server to another.
These elements help contribute to a reduction in footprint and reduced power and cooling cost. IBM's top-of-the-line server – the p595 is actually built on a Mainframe chassis – and provides many of the same reliability, availability and serviceability (RAS) benefits of the IBM mainframe. Using Linux in a data center migration strategy provides another important advantage: market product sustainability. Linux is the only OS that is actually growing market share, which will help companies protect the investment which they would be making. Further, most of today's innovation is occurring with open source systems like Linux.
In my opinion, one of the best data center consolidation strategies you can use is building around Linux. Data center consolidation projects allow business to cut costs, increase efficiencies and improve upon the ability of IT to support the business. Consolidating operating systems as part of a greater IT strategic plan will help you make optimum use of your data centers and the personal supporting your infrastructure.
ABOUT THE AUTHOR: Ken Milberg is a systems consultant with two decades of experience working with Unix and Linux systems. He is a SearchEnterpriseLinux.com Ask the Experts advisor and columnist.
