Co-author: Meyyappan Ganesan

People have been predicting the end of mainframe technology for decades. Unfazed by the naysayers, mainframes, the work horse of large enterprises have been quietly chugging away tons of computing workload for decades, without much fanfare. IBM sales figures indicate that mainframes are thriving and not just surviving. Mainframes typically handle large workloads, be it data processing, transaction processing, or being the systems of record. The larger the workload, the more cost effective they are. At high workloads they become the cheapest, most reliable, and most available system out there. In addition, the mainframe technology typically runs at close to 100% CPU efficiency whereas most intel-based server farms perform below 60% utilization, which means better returns on the capital expenditure.

IBM sales figures indicate that mainframes are thriving and not just surviving.

History and adapting newer trends

IBM introduced the large-scale computers in the 1950s with the 700 series of machines. There were series of developments in hardware and software, post this.

COBOL and Fortran got added as programming languages. With System 360 came time sharing and multiple virtual storage. Later came other programming languages such as PL1 and C. In the 1970s, the new ESA version of the machine extended addressability to 31 bits. Parallel sysplex and data sharing were introduced resulting in better load balancing and 24X7 availability and fail over.

In the 2000s, Java and Linux followed by native webservices and 64-bit processing saw the light. Post 2010 saw introduction of support for APIs and REST services, hardware enhancements for Java, C++ workloads and hardware based pervasive encryption. IBM and other ISVs such as CA and BMC continued to invest heavily on this platform with new cloud-based products and open frameworks called Zowe. IBM has adopted continuous delivery paradigm to roll out its products to the market. With the introduction of Red Hat OpenShift on z-HW this year, hybrid cloud and some share of frontend applications moving onto z-servers are a reality.

The current mainframe modernization footprint

These innovations ensured mainframes are the first choice for enterprises looking for throughput and reliability. There is an estimated 220 billion lines of COBOL code, besides code in other esoteric languages such as PL/1, Assembler, and Natural etc. As per industry reports, 92 of the top 100 banks worldwide and 71% of the Fortune 500 companies run their core business on mainframes. Mainframes host more transactions than Google (1.3 million/sec vs 68,542/sec for Google) including 55% of all enterprise transactions.

In addition, IBM has reported a ~ 60% increase in mainframe sales in the last three quarters owing to the release of z15 HW launched last September (https://www.ibm.com/investor/events/ibm-2q-2020-earnings-announcement - Server growth 6%, z Mainframe 69%).

Though IBM dominates the mainframe market share, other vendors such as Tandem, Fujitsu, Siemens, Unisys and Hitachi, NEC etc., manufacture mainframes but their market share is considerably less and GEO specific (~10%).

Market Segmentation and Trends

Let us look at how this market is segmented. As per a Gartner study, there are four segments based on the processing size (MIPS) used.

• < 5000 MIPS - Small users - 300 to 400 customers
• 5000 – 20000 MIPS - Medium users - 1,100 to 1,700 customers
• 20000 – 75000 MIPS - Large users - 1,800 to 2,500 customers
• < 75000 MIPS - XL users - 250 to 300 customers

The growth in these segments follow a specific path. The small size customers are moving out of mainframe at about 5% rate. The medium size customer is growing at 5%, large ones from 5 to 15%, and the XL customers are increasing their mainframe footprint from 15 to 20%. Gartner predicts that over the next five years, the overall mainframe installed capacity is going to grow about 6 to 8% annually with financial, public sector, and manufacturing growing more than the other verticals.

This is in alignment with what we see with our customer base and mainframe market.

Small customers opting to move to a shared data center or out of mainframe tend to adopt rehosting (move to a like environment) or migrate out.

The medium sized customers are choosing to keep their strategic applications on the mainframe. Moving out means long risky rewrite programs which may not provide the ROI expected. Hence, they are devoting their time in MIPS optimization and other cost optimizations and in-place modernization efforts. This means a reduction of yearly license cost and at the same time modernization of the applications and processes. It also entails moving non-core application out of mainframe and further migration to cloud.

The large sized customers are investing more on mainframe. They want to extend their core applications, consolidate their application space, and improve their process by adding DevOps and other automation while exploring cloud options. They have definite plans for their mainframe workloads and mainframe continues to be the core of their enterprise landscape and technology roadmap.

In effect mainframe surround is the long-term strategy adopted by medium and large sized customers. They will continue to optimize, modernize, consolidate, and trim their mainframe workloads and improve accessibility, which will increase the business value of their core applications. Meanwhile, IBM continues to implement latest advances such as native API support, open source integration, analytics, and other cost reducing and performance enhancing upgrades.

Mainframe surround is the long-term strategy adopted by medium and large sized customers.

How customers are thinking about mainframes

Mainframe customers are looking for -

• Reduced cost of ownership
• Improved time to market for enhancements.
• Integrate the overall software processes across the enterprise landscape.
• Creating an ecosystem where mainframe applications collaborate with the other technologies and channels, meaning that business functionality and data is rationalized across the entire landscape.

An aging work force is another critical challenge faced by customers. There is a dearth of skill pool capable of replacing resources who designed and supported business critical applications. To address this issue, many organizations, HCL being one of them, continue to invest in training new talents on the platform. In addition, new open source platforms called Zowe are being designed to appeal to new generation programmers.

HCL has a set of targeted services in addition to ADSM to cater to the varied requirements of our customers.

• Revitalize - Enhance the current platform and increase the value of the assets. Treatments include: performance optimization, In-place modernization, technology upgrade, and offload
• Rebuild - Rewrite the application from ground-up by extracting rules and re-designing (re-engineering)
• Convert/Port - Tool based conversion to target technology stack (Java, .Net). Rehost to like environment, Port to LUW by reusing COBOL
• COTS/SaaS - Migrate to industry specific products with functionality match
• Cloud Services - Move to cloud native platforms including IBM zCloud

HCL’s Mainframe practice has more than 25 years of experience in delivering mainframe projects across multiple verticals - application support and maintenance, development, consulting, performance optimization, migration, modernization, and product engineering. HCL has built several tools and solutions for mainframe discovery and modernization services (iLIT DC Suite) which is used in customer engagements, in addition to partner solutions.

Conclusion

Mainframe platform is here to stay. As per IBM sales figures, the platform is thriving with more processing power added. IBM continues to enhance the platform with new technologies and innovations such as cloud, DevOps, Open source software platform etc. Besides, exiting mainframe is not a panacea in every situation. Understanding the customers’ needs and providing a tailored approach with due appreciation of their business needs and technical roadmap is the right approach.