Data Center Tech Blog

The Software Mainframe leverages lots of virtualization – and flash. Jean S. Bozman looks at how software-defined architectures turn data center silos to grids.

When Paul Maritz became CEO of VMware in 2009, he said the company would be building “the software mainframe” by virtualizing computing resources in the datacenter—and managing them as a holistic system.

I was in the audience, attending a VMware analyst meeting that day, and I was listening closely to his speech, but couldn’t see why he chose that analogy to a mainframe when he was explaining how VMcenter management software worked—and added value to the then-hypervisor-centered VMware software stack.

Today, in 2015, his vision of the software mainframe appears to have been prophetic because it is taking shape by harnessing the virtualized resources within the data center.

Flash and the Mainframe-Inspired Vision for a Virtualized Data Center

Flash has emerged as an important enabler of this mainframe-inspired vision, because high I/O rates in the virtualized infrastructure require high performance to make that style of data center computing a reality.

Through repeated testing, SanDisk® has documented how flash technology accelerates virtualized workloads. You can learn more about it here.

We see the entire industry buzzing about software-defined data centers, and software-defined everything and anything. It is the high level of abstraction (from the underlying hardware), the deep virtualization, the software management and the high-speed I/O in the infrastructure that makes it all possible.

Why the analogy to a mainframe?

Speaking of mainframes – the mainframe systems in the analogy applies most closely to IBM mainframes, which are the primary brand of mainframes in the world today – and the only ones that run IBM-compatible systems software. The IBM mainframes were the first systems, back in the 1960s, to become heavily virtualized – and it took decades until the move to virtualize server hardware accelerated, first in Unix systems and then in x86 server systems.

The IBM mainframes are dedicated, high-end systems that are highly optimized via hardware and software to do heavy transaction processing (OLTP), to support scalable databases, and to run virtualized workloads that run across the partitions of IBM’s mainframe hardware and their associated systems-software platforms.

SDDC Enters the Picture

But the analogy to these famously fast mainframe systems rings true in the software-defined data center (SDDC) world. Technology refresh is aimed at transforming the siloed data center into the seamless data center – with workloads able to move around the virtualized platforms with ease.

Today, VMware, Microsoft’s Hyper-V, Citrix’ virtualization for servers – all have brought deeper virtualization to the entire data center.

The trend began in 2005, when VMware’s hypervisor had a first-mover advantage in virtualizing large numbers of x86 servers running Microsoft Windows and Linux distributions. Since then, other hypervisors have gained share, with Microsoft Hyper-V now the second most widely used virtualization software for data center servers – causing competition in the virtualization space to grow.

How is that happening? It’s all about the power of virtualization, combined with fast server platforms and advanced software management capabilities.

How It Works — (or, How IT Works . . . )

Let’s take a look at what’s taking shape in the datacenter that reflects the main features of mainframe-style computing, including:

  • Virtualization of all computing elements and storage.
  • High-speed I/O within the virtualized systems.
  • Emergence of a virtualized network, with high-speed connections.
  • Management across the system, and software automation.
  • Strong support for IT policies, including security, SLAs and QoS.

But the notion of a software mainframe is being built on the foundation of software-defined data center (SDDC) technology. This technology virtualizes all the elements of a data center: compute, storage and networking – and ties it all together with software automation and high-speed “fabric” links across the network.

What’s Ahead for the Data Center? Silos Turn to Grids

Where IT silos used to bottle up applications within data-center domains centered on specific operating systems, or hardware types, the increasing standardization of the data center infrastructure is finally allowing workloads to flow across the data center landscape. Now, workloads are “mapped” to appropriate resources, either on-prem in enterprise data centers or off-prem in cloud data centers. This ability to move workloads to available resources is another mainframe-like capability—and one that is enhanced through software automation to get the workloads to the right platforms, SLAs and QoS statistics, depending on the workload requirements.

IT efficiency is improved, with IT simplification and workload consolidation creating “pools” of compute and storage resources. Flash technology – including SSDs, PCIe and flash-optimized software (e.g. SanDisk ZetaScale), has an important role in all of this.

Flash allows the storage resources to keep pace with faster network connections, and faster processors. One example is SanDisk’s certification for VMware VSAN configurations, which support software-defined storage using flash. You can read more here.

This IT efficiency translates into real-world operation cost reductions for the businesses that depend on those data centers – or buy data services on an hourly basis from the cloud providers.

Planning for the IT Time Horizon

But IT managers must plan, working in cooperation with business planners, so that company resources are spent on efficient computing platforms, rather than on modernizing 30-year-old COBOL code. Typical IT time horizons are 1-year, 3-year and 5-year plans, allowing for long-range planning as the infrastructure is transformed, over time. They must plan to maintain service level agreements (SLAs), and quality of service (QoS) levels that the business has already established.

Sure, it’s true that more work needs to be done before the inexorable forces of technology refresh, combined with software-based virtualization, make Maritz’ vision of the “software mainframe” become a reality in enterprise and cloud data centers.

These data-center-wide systems will have to become more expandable, and flexible, and secure. As this happens, flash-enabled servers and flash-enabled storage will become more widely installed throughout the network, because faster systems impact the “take-home” results and the financial “bottom line.”

But—make no mistake about it – the march to build and deploy more “software mainframes,” in the form of software-defined data centers,  has already begun. And, it will happen in stages, as highly virtualized data centers are increasingly implemented. While “software mainframes” are not clones of the world’s physical mainframes, the move to SDDCs is a powerful idea aimed at unifying the IT platform for application deployment.

These 5 Forces Shape Our 2015 Technology Trends

A First Look at the Future During VMware Partner Exchange

Subscribe Today!

Get our latest posts via email and enjoy data center insights in a flash