Authors: Ankush Rathore, Anurag Vij, and Francois Magnin, Azure Cloud & AI Practice, Microsoft Industry Solutions

“After two failed migrations over seven years, ‘mainframe migration’ became the two most dreaded words in the organization. Third attempt was not only opposed by the business unit heads but was outright rejected by the steering committee…” – Group CIO, A Regional Bank


Mainframes figure prominently in the history of computing and remain viable for highly specific workloads. They continue to be used across industries to run vital information systems, and have had a place in highly specific scenarios, such as large, high-volume, transaction-intensive IT environments.

The typical mainframe operations are:

·      Online: Workloads include transaction processing and database management.

·      Batch: Jobs that run without user interaction, and typically on a regular schedule.

·      Job control language (JCL): Specify resources needed to process batch jobs.

·      Initial program load (IPL): IPLs are used to recover from downtime. An IPL is like booting the operating system on Windows VMs.

A system that was first designed in the late 1950s as scale-up servers to run high-volume online transactions and batch processing, continues to be an Achilles-heel for big enterprises. It is the last and biggest hurdle that holds organizations back from making 100% move to the cloud.


The reliability, availability, and processing power of mainframes have taken on almost mythical proportions, and the below table should help in distinguishing the myths from the reality (1).

No alt text provided for this image


The Mainframe has stood as the high-powered safehouse of critical enterprise applications for decades. For many businesses, it remains at the very heart of many organizations’ core line-of-business applications and workflows and viewed as a symbol resilience.

Mainframes have been the proven platform that scale well and provide reliable performance, assuming there are programmers and developers available to design, run, and maintain mainframe programs. In a robust Mainframe operating environment with long-established operating procedures, including JCLs (Job Control Language statements), programs can run based on usage, measured by MIPS (million instructions per second), and provide extensive usage reports for charge backs.

However, rapidly evolving cloud technologies that provide significant business, economical, operating, and personnel advantages over Mainframes build a solid case for organizations to seek a mainframe-free state.

Business Value: Enterprises unlock business value for themselves and their customers by driving continuous innovation. Innovation in a digital world requires a move to a modern DevOps environment. Mainframes are typically monolithic platforms where workloads and surrounding processes don’t necessarily lend themselves to enabling agile development or rapid innovation cycles. Move to cloud platform enables organizations’ IT and business units to closely partner to create opportunities for innovation and supportive services that drive transformation.

Azure as a leading cloud platform provides a hyperscale environment for mission critical workloads promising multiple 9s of availability, optimized for local or geo-based replication services, and backed by commitment-based service level agreements (SLAs). Moving to Azure or similar cloud-based platforms provides organizations the ability to innovate faster, develop and operate better, and consequentially, unlock business value faster to stay ahead of the competition.

Economic Value: Pandemic has had a multi-fold acceleration impact on digital transformation of organizations across industries. In the current environment, organizations are pushed to innovate their future businesses at unprecedented rates while they survive intense competition and operate in the present. Thereby, every dollar saved is an opportune extra dollar to create economic headroom for faster transformation.

The perception of mainframes providing operational stability, and thereby cost savings for the organization is an anti-pattern and must be avoided. To start with, mainframes are subject to very expensive monthly hardware and software contracts, where extra capacity comes at an added cost while not providing the elasticity that Cloud services can. There are several studies and examples of a 5,000 MIPS environment being nearly tenfold more expensive than leveraging Cloud services such as Azure.

The operating and maintenance cost of mainframe environments increase by 4% – 7%, annually (source: BMC). Lack of pay-for-use models, CAPEX and OPEX commitments, and rising costs to source an ever-decreasing workforce with mainframe technical and developmental expertise further makes it a losing proposition.

Lastly, and perhaps most importantly, a mainframe environment stifles rapid experimentation, learning, and innovation.

Personnel Considerations: Industry has seen a constant attrition of mainframe knowledge over the years. Like a prominent CIO noted,

Even if I wanted to run mainframes, lack of people available with quality technical skills and abilities to manage them makes it almost impossible for me to do so.

The decreasing number of mainframe developers that can develop and maintain mainframe applications not just makes it more expensive to operate these environments but also creates an environment that’s change averse and hinders creation of new value on existing mainframe environments. In contrast, modern mainframe-free operating environments with large pool of deep technical skills help remove data siloes, help create new value within the organization and its ecosystem, and helps businesses compete better while making them future-proof.


Once an organization has understood the trade-offs, making migration decisions must be taken with thoughtfulness and planning. The lifetime of mainframes within the organizations typically exceed their oldest employees. This has made mainframes deeply embedded into many organizations’ core business processes and staff’s way of working. As deeply entrenched as this has made mainframes both technically and process-wise, it has also created what’s commonly termed as technical debt and process debt. Whilst all organizations carry a certain amount of technical debt and process debt from the past and must consider that in their business cases as they plan modernization of their environments, with mainframes this debt can be multifold and hence requires a thorough and well-bought-in case for change. There are several other factors such as technological choices, project management, and change management that must be considered and planned for too.

The Case for Change: Typically, mainframe migrations are driven by the central IT teams. Levels of reluctance is usually encountered from the business units, where such a large change not only requires considerable incremental (and often undesired) efforts and resources from them but also poses a threat to their businesses and priorities, should things not go as planned. A lack of strong buy-in and ongoing commitment from key stakeholders that are crucial to provide critical inputs during planning, implementation, and adoption of new systems during a mainframe migration stays a top reason for why mainframe migrations fail. The case for change must be strong with complete stakeholder buy-in and ongoing support as a critical dependency for success.

Technical and Process Debt: Myriad of applications, business processes, and mere code that’s been created over decades from simple patches to JCLs have succeeded in finding deep roots within the organization. Untangling this, technically and process-wise requires deep discovery of applications across the portfolio as well as identification of technical and process dependencies. Depending on the industry the organization is operating in, this could also mean looking at regulatory or other compliance requirements across the countries that the organization operates in. Many mainframe migrations fail due to lack of deep discovery and mapping exercise during planning, only to be discovered during execution, and in many (unfortunate) instances requiring a complete rollback.

Technological Choices: As planning teams carry out thorough discovery activities, its critical to understand the application’s purpose, business rules, underlying code, available documentation and other support resources, and its technical dependencies. These factors help determine the treatment these applications should be given during the migration process ranging from Re-platforming, Re-factoring, or Re-architecting to Retiring (and repurchasing). Post migration planned steps and strategies to test and mitigate issues that could concern scale or redundancy of these applications must also be a key part of discovery and planning process. This requires right capabilities, access to application owners, and a well thought out planning process. Given the complexity, many organizations lead with a lift-and-shift or Re-platforming approach, wanting to move the applications as-is to Cloud services. This not only leads to many failed mainframe migrations but also mostly defeats the purpose of moving to Cloud, technically and economically.

There are several possible approaches to a mainframe migration depending on the application stack and the underlying legacy hardware it sits on. These range from emulating the hardware in Cloud while the stack is ported as-is followed by treatment of middleware, application code, data, and so on to treatment of the stack followed by porting to Cloud services. Many of these applications may be treated differently while they inherit the same databases. A key factor to determine during the process is critical data dependencies that keep the business functions and required analytics and reporting intact. Failure to do so impacts core business functions and results in unfortunate cases of rollbacks.

Project and Change Management: Research proves over 60% of the projects either completely fail or fail to achieve the originally anticipated outcomes. Apparently, the percentage of failure for mainframe projects is even higher (2). Mainframe projects require deep levels of project management. Where most mainframe projects fail, two key areas are either underestimated or left unaddressed: (a) deep levels of project management embedded within the application teams that ties back into the larger project framework, and (b) change management program that runs in parallel to drive acceptance, readiness, and plan-to-scale of the application teams. Lack of attention to these areas are a definite way to fail a mainframe migration.


Microsoft and its partner ecosystem take pride in helping hundreds of customers with their mainframe migrations and accelerating their digital transformation journeys. Microsoft Industry Solutions believes deeply in a migration framework that’s embedded in core principles of: (a) Discovery and Assessment, (b) Building a strong business case, (c) Making smart technological choices leveraging Microsoft’s deep partner ecosystem, and (d) Leading with world-class project and change management.

Discovery and Assessment: Based on the key business drivers that a client has set to modernize their mainframe workloads, Microsoft Industry Solutions (IS) conducts a Discovery and Assessment phase leveraging one or more of its partners and leveraging its global intellectual property to determine the best migration strategy. Given there is no one-size-fits-all for modernizing Mainframe workloads, this uniquely crafted migration strategy serves the specific business outcome that the organization is seeking for within the requisite timelines.

Building a Business Case: The Business Value Advisors (BVAs) work closely with the client to help build a robust business case that helps the client to seek internal buy-ins and form stakeholder coalitions. In addition to requisite financial approvals, the business case helps bring organizational stakeholders, from Boards to Business Unit Heads, along in the journey from the get-go.

Making Technological Choices: Based on the migration strategy, IS team works with its key partners and client’s stakeholders to determine and finalize the technological choices. These choices are often determined through simple to complex POC, depending on the application or workloads under treatment. POCs not only validate the underlying assumptions and expected outcomes, but also help determine mitigation strategies for anti-patterns, code gaps, and scale challenges. Depending on the technological choice, automated tools such as AI based code reverse engineering are often used to deconstruct mainframe applications, to achieve speed and scale.

World-class Project and Change Management: IS firmly believes that both strong project management and change management are fundamental to driving any large migration or transformation effort. It’s equally, if not more, important to mainframe migrations too. Embedding project management into application teams aligning them closely with application owners assists in driving a predictable project that can actively discover and mitigate forward looking issues and avoid unwanted surprises to the overall program. Driving a thorough adaption and change management program assists in building user awareness, securing buy-in, building user readiness, and creating scale abilities.


While mainframe migrations can be a daunting task, whether an organization is looking at becoming mainframe-free or has a path to operating in a hybrid environment while growing its MIPS for certain workloads, the outlined underpinning strategies to a successful migration project are core to success. A business case that’s built upon a well discovered and assessed environment, helps secure stakeholder buy-in and support, informs right technological choices, and is executed upon with world-class project and change management offers highest possibilities of success.