EEveryone with a data center has at least one common goal: making or saving money for the company. They're trying to lower total cost of ownership (TCO), increase business agility and simplify the complexity. One obvious way to do this is with server virtualization; that is, consolidating workloads running on several servers onto one generally more powerful server. Once you agree that server virtualization is a great solution, you start wondering which workloads to consolidate and what virtualization infrastructure to use.
Often, those who accept server virtualization as a solution start the virtualization process by selecting a hypervisor and then the work of consolidating workloads running on physical servers onto virtual machines created by the hypervisor. Don't take this approach. Rather, start with an analysis of your data center workloads to determine which are good candidates for consolidating.
Eventually, you'll have to select virtualization software to create your virtualization environment. Most of the options are x86-based. Yet, another choice-consolidating workloads onto mainframes -is often overlooked, but now is becoming more popular. Often, you'll find it can be the most cost-effective approach to server virtualization because mainframes aren't as expensive as they previously were, and they've gotten much more powerful over the years. In addition, they require a fraction of the floor space and power that comparable-capacity x86 servers need. And you get the benefits of a mainframe such as increased hardware reliability, extreme I/O throughput, and less downtime. Those benefits just come with mainframes and you won't automatically get them from server virtualization on x86 servers or any other architectures.
To use Linux on a mainframe as a server virtualization technology, make sure the workloads that you'll consolidate run on Linux or at least can be ported to Linux.
Two flavors of commercial Linux distributions run on mainframes: SUSE Linux Enterprise Server for System z from Novell and Red Hat Enterprise Linux for System z for Red Hat. You choose which to use, but know that Novell has an 80+ percent market share.
Virtualization with Linux on the Mainframe: How It Works
Mainframes are about 40 years old. And as you probably know, several years ago mainframes were being written off as expensive, dated computers that were capable of running only large business applications. Not anymore. Mainframes have evolved rapidly during the past few years with Linux and innovation from IBM. But, innovation has maintained the benefits of mainframes that have always made them desirable:
- high utilization rates: 80-95 percent over a 24-hour period
- ability to run mixed workloads without user intervention
- efficient use of floor space and power
- virtually zero downtime
Today, mainframes are faster, smaller, and far less expensive than their ancestors. Take for example, the IBM System z10 Business Class server. The z10 BC gives you usable capacity equivalent to 232 x86-based servers, while using 83 percent less floor space, and up to 93 percent less energy. The cost per millions of instructions executed per second (MIPS) is one way that IBM compares the cost of mainframes. Today, the cost per MIPS for the System z10 is a very small fraction of the cost per MIPS for mainframes sold in 1980. This means that you get significantly more speed and capacity for your dollar than with older mainframes. You can get an IBM z10 BC for around $100,000, and it gives you significant capacity with its new 3.7GHz quad-core processors. Plus, you get all the benefits of newer, more innovative technologies. Not bad at all when compared to the specs and pricing of your grandpa's mainframes.
Creating virtual machines to run Linux workloads on the mainframe begins with z/VM, IBM's mainframe hypervisor operating system. z/VM runs in a logical partition (LPAR). When it creates Linux virtual machines, it shares CPUs, memory and network resources among the Linux virtual machines. It also manages Linux virtual machines. Linux virtual machines created using z/VM are run using standard mainframe processors, also known as Cps, or an Integrated Facility for Linux (IFL). An IFL is a specialty engine that IBM created to make running Linux-based software more economical. For example, when you buy software, say an expensive app such as an Oracle database, Cps and IFLs are treated as a single processor, so the app can be used
Subscribe to Connection Magazine
relatively inexpensively by all the workloads running in the Linux virtual machines. (See Figure 1.) More on that later.
To use Linux under z/VM, you'll need to buy enough CPU and memory capacity, and a Linux for mainframe subscription from Novell or Red Hat. With the use of z/VM functionality, you can begin creating Linux virtual machines in minutes and hours compared to days or weeks for distributed hardware. Today, there are more than 4,000 IFLs installed and more than 1,300 customers running Linux on the mainframe.
Linux on mainframe distributions aren't much different than their x86-based counterparts. Differences involve the implementation of features that take advantage of mainframe hardware and mainframe operating systems such as z/VM. These features aren't generally visible to application developers so they don't need much, if any, training to be productive.
With Linux on mainframe distributions you can generally run the same applications that run on Linux on x86-based architectures. Of course, this requires software vendors to support the architecture, which not all do. If your vendor says that they have a version of their product for Linux, make sure that they have one for Linux on System z.
Novell makes it easy for you to try out SUSE Linux Enterprise Server for System z, even the new release 11, on a mainframe for free. Just download SUSE Linux Enterprise Server Starter System for System z and install it in less than 90 minutes. If you don't have the z/VM operating system running on your mainframe, you can get a free eval from IBM. The Starter System doesn't require much experience with Linux and it gives you the chance to do proof of concept evaluations for free.
Reasons to Consider Linux on the Mainframe for Server Virtualization
TCO for using Linux on mainframes includes the costs of:
- software licenses
- floor space
- system management tools
- system administrators
- network hardware (and administrators)
- backup and restores
- and so on.
Following are several reasons why Linux on the mainframe is an excellent way to meet your data center challenges and reduce costs.