|
|
Illustration by Peter Bollinger
Virtualization is a key enablement technology for the New Enterprise Data Center, but not all servers are created equal when it comes to virtualization capabilities. This is because server virtualization support isn’t an “add on once and you are done” investment proposition. A platform’s virtual-server hosting capabilities are greatly influenced by the underlying system architecture, hardware and firmware even before you consider the breadth of function offered by the hypervisor technology available on the server.
The IBM System z10* platform, the latest generation of mainframe technology, has a rich heritage of innovation in the area of virtualization. That should be no surprise considering the fundamental “share everything, host anything” architecture of mainframe computing. Equally noteworthy is the fact that the most demanding customer set in the world, mainframe users, has been using this platform for business-critical workloads for decades. This iterative process of investment, utilization, feedback and more investment over many years has helped IBM strengthen its mainframe virtualization technology, offering exceptional levels of resource utilization, workload management, transaction and data integrity, operational efficiencies and business resiliency. For some, talking about the rich history of mainframe-virtualization support seems almost clichŽ, as if mainframe innovation is a thing of the past, with no opportunity for future technology advances.
With a nod to the IBM System z* family as the leader in virtualization, many distributed system vendors speak of mainframe-like technology in their product offerings. This is an interesting development considering, in the not-too-distant past, these vendors took great pride in touting the architectural differences of their systems as compared to IBM mainframes. It seemed, at the time, the prospect of installing yet another box to grow application capacity was a virtue, offering the autonomy many business units desired while leaving the care and feeding of the growing server sprawl to IT staff. More than just a code of conduct for the hardware vendors, the whole distributed-server ecosystem embraced this style of application hosting, which one could summarize as costly independence. It’s a fairly recent awakening in the IT industry that this is no longer a viable model for computing, yet serious new issues need to be addressed when one begins to move away from this style of computing. The problem is so much of the distributed-system computing landscape has been built over the years on a foundation of operating-system independence with dedicated system assets.
To maximize the efficiency of a New Enterprise Data Center, it’s important for IT executives to understand why the System z platform is considered an industry leader with its virtualization support. It’s especially imperative for existing System z clients to understand the particular attractiveness of deploying virtualization solutions on their existing set of mainframe assets.
Mainframe virtualization support has been an essential element of its architecture since its inception in the 1960s. Over the course of many years, IBM has refined the System z infrastructure with a coordinated investment in hardware, firmware, hypervisors and operating systems—to enable exceptional qualities of service for hosting virtual servers. When clients look for enhancements in the areas of speed, scalability, resource efficiency and security, IBM has introduced function in the proper technology layer, or set of layers, to provide the desired results.
It’s unnecessary to insert code into the hypervisor layer when it makes more sense to update the firmware and operating system layers instead, or vice versa. For example, the integrity and security of a z/VM* virtual machine is provided by storage-protection keys in the hardware, instruction isolation in the firmware and privileged z/VM hypervisor interfaces for external security managers like Resource Access Control Facility, just to name a few of the components, that help secure a large-scale guest environment.
Coordination of virtualization enablement extends to other areas as well. System z firmware allows Linux* virtual machines to execute storage-area network (SAN) and Ethernet I/O operations without requiring the z/VM hypervisor to translate virtual data addresses; the firmware understands z/VM control-block structures and is able to perform the I/O operation without having to interrupt z/VM, enhancing I/O performance and overall system throughput. Similarly, in the management of virtual and real memory, firmware assists allow z/VM and Linux guests to efficiently manage the allocation of real memory so overall system performance can be optimized. Operationally, the System z Hardware Management Console exploits a set of z/VM programming interfaces to provide users a graphical interface to perform basic z/VM system management tasks without requiring an up-and-running TCP/IP network.
A multidecade, system-level investment focus is one reason the System z platform is in a leadership position for virtualization support.
A multidecade, system-level investment focus is one reason the System Z platform is in a leadership position for virtualization support
Page 1 2
Browse products and services for Trends.