| Mainframe Virtualization and Cloud Computing - State of the Software |
|
|
|
||
|
|||
|
Mainframe Virtualization
|
Virtualization plays an important role in mainframe hardware. By implementing virtualization software, it is possible to virtualize the processor, memory, storage I/O and networking resources into aggregated pools - applications can utilize resource pools without regard to specific hardware. This improves utilization of the mainframe, and reduces the need for upgrades. Mainframe hardware typically experiences degradation in performance when a greater number of users run applications that are increasingly complex. Common upgrades to IBM mainframe systems include improvements to processing power and memory. New processors help to tackle more calculation-intensive tasks, additional memory supports more sophisticated applications with less storage access, and cryptographic coprocessors handle encryption and decryption dynamically. |
 |
Other compelling reasons for mainframe upgrades are to improve processing for virtualized operating systems and development platforms: Solaris, Linux, DB2 for z/OS, and Java. IBM is modernizing the mainframe platform. For example, the z/VM operating system can support hundreds of Linux images simultaneously.
Web-enabled applications can be accessed by the company web site or a native application. An advantage associated with a web-enabled application is the utilization of other web services such as security and personalization. Web service information can help administrators troubleshoot mainframe applications.
Although too early to definitely predict an industry trend, it would appear that a number of organizations are in the initial stages of evaluating whether mission critical applications can be moved from distributed Intel servers to virtualized UNIX-variant environments on IBM System z mainframe systems. The economics underlying the cost savings are a reduction in data center real estate, middleware and application licensing costs, and the cost associated with volume purchases of Intel-based servers to meet growing capacity requirements. There also is the issue of the increasingly prohibitive data center power and cooling requirements which may ultimately shift x86 workloads to IBM mainframe systems. Kilowatts per server are now starting to be measured in the acquisition costs of Intel servers.
Mainframe Virtualization Technologies
IBM System z advanced hardware provides the capability to logically partition the machine, share CPU, memory and I/O channels and associated devices, add and remove computing Capacity on Demand, and provide high-speed communications among partitions. System z hypervisor technologies collectively provide the capability efficiently to support and dispatch multiple LPARs: Logical Partitions and virtual machines. PR/SM: Processor Resource/Systems Manager is System z hardware technology using a Type 1 hypervisor that allows multiple operating systems to run on the same physical processor, with each operating system running in its own LPAR. The Type 1 hypervisors run directly on the system hardware. Type 2 hypervisors run on a host operating system that provides virtualization services, such as I/O device support and memory management.
System z PR/SM can currently support up to 60 logical partitions.
z/VM runs in an LPAR and provides a virtualization layer designed to allow the capability to run hundreds to thousands of virtual server images for larger deployments.
There are two types of partitions:
|
Dedicated |
A dedicated partition runs on the same dedicated physical processors at all times and the processors are not available for use by other partitions even if the operating system running on that partition has no work to do. This eliminates the need for PR/SM to get involved with swapping out one guest and dispatching another. |
| Shared |
Shared partitions can run on all remaining processors that are not be being used by dedicated partitions. This allows idle systems to be replaced by systems with work to be performed at the expense of PR/SM overhead incurred by the dispatching of the operating systems. Shared partitions provide increased processor utilization, but at the potential expense of performance for a single operating system. |
Cloud First - United States Federal Government
On-line demand computing is the new model for improving ROI: Return on Investment on information technology hardware. Cloud computing has been a beneficiary of virtualization enabling many organizations to share computing resources and pay only for use. As part of the Cloud First initiatives, federal agencies are required to consider cloud options before new investments are made in information technology. The government also is consolidating servers and data centers in an effort streamline its costs. Significant benefits associated with virtualization are resource consolidation: less equipment, less space, and reduced consumption of electricity. A major advantages associated with virtualization is that data can reside across a shared pool of storage devices, but the devices themselves don’t have to be equal. Critical information that needs to be accessed frequently can be sent to high-performance storage - the equipment with the fastest response times - while less important data can go to lower cost devices with slower response time. Data that is rarely accessed or needed only in emergencies can be stored on devices that are less advanced, and less costly.
Cloud Computing and IBM
Reflecting its growing business in cloud computing, IBM has opened a new data center in Songdo International City, Incheon, South Korea, and announced plans for others in Auckland, New Zealand and Raleigh, North Carolina as part of a three-year program to upgrade and expand IBM’s world-wide network of data centers. This upgrade cycle is being driven by a growing shift by organizational enterprise to cloud computing. The trend is clear, data center operators are concerned about the amount of electricity it takes to run computers and keep them cool. Firms are increasingly interested in moving computing tasks to data centers operated by providers like IBM and Hewlett-Packard’s EDS operating unit. IBM also has announced that it will be opening an 80-person research laboratory in Hong Kong to develop collaboration technology for cloud-computing applications based on IBM's Lotus software. It will be an extension of IBM’s 5,000-person China Development Laboratory.
