Despite many years of predictions that mainframes would pass from the scene, enterprise workhorses such as the IBM Z and IBM i are not only thriving...
Despite many years of predictions that mainframes would pass from the scene, enterprise workhorses such as the IBM Z and IBM i are not only thriving but also still crunching 90% of all credit card purchases and running 68% of production workloads, according to the Share user group.
What’s more, the need for the mainframe’s highly available and resource-optimized architecture shows no signs of abating. Of
the 200 IT executives surveyed in 2021 by the IBM Institute for Business Value and Oxford on this topic, 82% agreed that their
business case supports mainframe-based applications, 74% said mainframe-based applications are central to their technology strategy, and 71% said they are central to their business strategy.
“The mainframe is the backbone of the world economy, not because of incumbency but because of the value the platform provides,” says Steve Nekolaichuk, a global offering manager for Core Enterprise and zCloud
However, organizations that rely on these sturdy platforms have felt pressure to
migrate to platforms that are perceived to be more modern. That would make sense if mainframes had stopped evolving 50 years ago. But as this e-book shows, “big iron” has not only kept pace with the speed of technology advancement but, in some cases, has also exceeded it by a wide margin.
Despite the overall robustness of the mainframe, perceptions remain that the value of the platform does not justify its cost. For example, the cost of a dedicated data center infrastructure can appear prohibitive and maintaining a mainframe is thought to require specialized knowledge from a supply of skilled professionals that is decreasing. It’s true that people with expertise in Job Control Language (JCL), 360 Assembler Language, and Virtual Storage Access Method (VSAM) storage are hard to come by. Although there are still Assembler Language and COBOL programs on the mainframe, the modern mainframe can also host cloud-native applications by using software containers, Java, and Linux. Some of the more traditional middleware and programs can be opened up for integration with standard application program interfaces (APIs).
“At nearly every level of the technology stack, the perceived complexity of the mainframe can be masked by open solutions and new platform features,” Nekolaichuk says.
As the world’s leading services provider, Kyndryl has the deep mainframe experience and expertise that can help enterprises transform while keeping their existing environments optimized. This can include providing mainframe infrastructure as a service (IaaS) as well as replatforming if that is required.
Another challenge organizations face
is the need to control the costs of hardware/software, facilities, and infrastructure management. A 2020 Deloitte survey revealed that the average enterprise spends 57% of its IT budget on business operations and only 16% on innovation.
Data center space limitations and competing priorities from new workloads such as lease expirations, consolidations, and relocations required by mergers and acquisitions also contribute to overhead. Kyndryl can help reduce or eliminate data center costs for its customers, by hosting their IBM Z and IBM i environments in its data centers.
The mainframe is often characterized as expensive, but when the total cost of ownership (TCO) is considered, it is, in many situations, the most cost-efficient option. Despite running 68% of production workloads, mainframes consume only 6.2% of worldwide IT spend, according to Share. The architecture is optimized for efficiency and high utilization. Whereas many servers run at an average of only 10% to 15% utilization, “the mainframe is architected to use all of its capacity,” Nekolaichuk says.
This is accomplished by having very efficient load balancing, parallel processing, and resource sharing built in. This is not only for z/OS but extends to Linux and other operating systems.
IT leaders have also recently been under
the gun to “modernize” applications, a dynamic that has been accelerated by the need to shift business models since the onset of the COVID-19 pandemic. The concept of modernization rests on the stereotype that
if mainframe hardware is out of date, the applications that run on it must be old and underpowered as well.
The first assumption is simply not true. A comparison of the mainframe in 1964 with the latest IBM Z release shows that performance, measured in millions of instructions per second (MIPS), has increased 531 million percent and onboard memory capacity has grown a staggering 63 billion percent. In both cases, mainframes have handily exceeded the Moore’s Law standard of a doubling of performance and capacity every two years.
To use an analogy, if automotive technology had moved ahead at the same pace, today’s cars would get five billion miles to the gallon and travel at the speed of light.
The second argument simply makes no
sense. Moore’s Law doesn’t apply to software applications that are doing the job they were intended to do. They don’t need to be rebuilt but, rather, can be extended with APIs, exposed as cloud services, and encapsulated in containers without alteration of the underlying code.
If auto technology had moved at the
same pace, today’s cars would get five
billion miles to the gallon and travel at
the speed of light.
Rearchitecting applications is time-consuming, expensive, and risky. In a survey by Advanced Computer Software Group, 78% of the responding organizations reported that they had started at least one modernization program as a result of the pandemic crisis and a nearly equal number said that at least one of those efforts had failed, mostly due to lack of planning and skills.
Kyndryl’s Nekolaichuk is more direct.
“The term modernize is often used by cloud providers as code for ‘Let us move everything you do to a new platform, rewrite your working applications, and charge you a lot
for it,’” he says. The best approach is to match workloads to the right platform, based on business and technical requirements. For many applications, the mainframe remains the right platform.
Companies that do want to restructure older applications typically move them to a Linux platform using cloud-native constructs such as containers. The good news for them is that IBM Z has been running Linux for more than 20 years, running containers natively, and supporting all the open tooling used in devops and other agile processes. By cohosting Linux workloads on the same mainframe with other Linux workloads, as well as with more traditional workloads, organizations can reduce network requirements and cut the latency between applications to near zero.
IDG Communications, Inc.