Mainframe modernization: What stays, what moves, and how to move it
One European bank cut the annual cost of its mainframe content by 54.1% within two years, and it didn't replace the mainframe to do it. The core applications kept running where they were. Only the documents moved.
That result points to the question most banks and insurers face with mainframe modernization. The core can't leave without a multi-year rewrite, yet the board wants progress this year. So the useful question is what to move first, and how to move it while the business keeps running.
Regulation weighs on every one of those choices. In Kyndryl's 2025 State of Mainframe Modernization survey, 94% of 500 business and IT leaders said regulatory compliance strongly shapes their modernization plans. This guide shows you how to split the work and where to start.
Key takeaways
- Mainframe modernization covers three paths: modernizing on the mainframe, integrating it with the cloud, and moving selected workloads to other platforms
- Most banks and insurers keep core transaction processing on the mainframe, and the data supports that choice
- Documents, archives and document workflows are the lowest-risk workload to move first, because the core keeps running untouched
- Content can move during live operations: new documents go to the new platform, metadata moves first and the archive follows in the background
- A complete record of every migration step is what lets you show auditors that nothing was lost or altered
What is mainframe modernization?
Mainframe modernization is the process of updating legacy mainframe applications, data and infrastructure so they meet current business demands. It ranges from improving applications on the mainframe itself to connecting them with cloud services or moving selected workloads to other platforms. For most enterprises, it is a phased program that runs for years.
The three paths to mainframe modernization
Kyndryl's research groups modernization into three paths, and respondents reported returns of 288% to 362% across them. The paths differ in what they change and how much risk they carry for a bank or insurer.
|
Path |
What changes |
Risk to daily operations |
Typical workloads |
|
Modernize on the mainframe |
Code, tooling and APIs on the existing platform |
Low |
Core transaction processing |
|
Integrate with the cloud |
New services connect to mainframe data and logic |
Medium |
Digital channels, analytics |
|
Move workloads off |
Selected workloads run on other platforms |
Depends on the workload |
Content, archives, reporting |
The third path is where mainframe migration happens. The risk depends almost entirely on what you choose to move, which is why the order matters so much.
Why banks and insurers keep their mainframes
Keeping the core on the mainframe is a rational decision. Mainframes process high transaction volumes reliably, and the business logic inside them took decades to build and test. Rewriting a core banking or policy administration system puts that logic at risk for a payoff years away.
The market data reflects this. According to the Kyndryl survey, fewer applications are moving off the mainframe than the year before, and 80% of organizations changed their modernization strategy in the past year. The shift is toward phased programs that deliver value without disrupting critical operations.
So for most institutions, the real planning work is deciding which workloads leave the mainframe and which stay. That is a hybrid model: the mainframe runs what it does best, and other platforms take on the rest.
Why manufacturing modernization keeps stalling on legacy IT
Discover how IT leaders in manufacturing connect that information first, giving modernization initiatives like AI the fuel it needs to deliver results.
Read nowWhat stays on the mainframe and what moves
The clearest split runs between transactions and content. Transactions need the mainframe's speed and reliability. Content needs low-cost storage and fast retrieval, and the mainframe provides both at a high price.
|
Workload |
Stays or moves |
Why |
|
Core transaction processing |
Stays |
Proven logic, high volume, low tolerance for change |
|
Batch posting and settlement |
Stays |
Tightly coupled to core systems |
|
Customer statements and correspondence |
Moves |
Grows every year and is read rarely, but must be found fast |
|
Claims files and loan files |
Moves |
Teams need them in their daily tools |
|
Report and print archives |
Moves |
Long retention periods drive storage cost |
|
Document workflows such as approvals |
Moves |
Easier to change on a modern platform |
The right column is where cost and frustration build up. Every statement and claim file your mainframe stores uses capacity you pay for, even when nobody opens it until an auditor asks.
Why documents should move first
Documents carry the least risk and show results the fastest. Moving them leaves your core applications untouched, so no transaction logic changes. Yet the effect shows up in your costs and at your front line.
The European bank behind the 54.1% figure faced three pressures at once. Storing and processing content on the mainframe drove its capacity-based costs up as volumes grew. Slow retrieval meant customer service teams kept customers waiting. And employees struggled to find data, which weakened reporting.
The bank moved its content management off the mainframe to a modern enterprise content management platform.
Annual costs fell by 54.1% by the second year, shortage fees dropped by 78.6%, and according to the case study, the change came with little to no downtime. That is one bank's result, but it shows how much of a mainframe bill can sit in content.
There is a second benefit. Once documents live on a platform built for content, they are ready for AI search and automated workflows. That gives your modernization program a visible win for business teams while the larger decisions about the core continue.
How to move documents off the mainframe without downtime
Hey Doxi, how can my team move documents off the mainframe without any downtime ?
1. Map your content and its retention rules
Start with an inventory of what the mainframe stores: document types, volumes and how long each must be kept. Record who uses each type and how fast they need it. This list decides your migration order and tells you which records carry the strictest retention rules.
2. File new documents in the new platform from day one
Stop adding to the problem first. Route new statements and claims documents straight to the new platform, so the mainframe archive stops growing. From this point, your teams find new documents in their daily tools.
3. Move the metadata first
Transfer the index data for every existing document before you move the files themselves. Your users can then search the whole archive in one place, even while the content still sits in its original storage. With the metadata in place, you can retire legacy systems far sooner than in a classic migration.
4. Migrate the content in the background
With the metadata in place, the documents follow in the background during normal operations. Users keep opening documents as usual and don't need to know where each file sits at that moment. For large archives with billions of documents, this approach shortens projects considerably.
5. Keep a record of every step
Each document should arrive unchanged, with every migration step logged and traceable. US record-keeping rules, such as SEC Rule 17a-4 for broker-dealers, expect you to show that records stayed complete and unaltered. A full migration log gives your auditors that evidence.
6. Retire the old storage
Once content and metadata are verified on the new platform, you can switch off the old storage and the capacity it consumed. This is where the cost savings become permanent.
Where mainframe content projects go wrong
Most failed content projects fail on planning. The method above works when you avoid three common mistakes.
Moving everything at once
A big-bang cutover concentrates all the risk on one weekend. A phased move, starting with the document types your teams use most, spreads that risk and shows results earlier.
Waiting until the experts retire
The people who understand your mainframe archive and its retention logic are leaving the workforce. Capture what they know about indexes and retention periods while they are still there to answer questions.
Skipping a format check
Mainframe archives hold legacy print streams, such as statements built from overlays and resources. Test how a sample of each format renders on the target platform early, before you schedule the bulk migration.
Modernize your mainframe content with Doxis
Your core applications can stay on the mainframe while your documents move to a platform built for them. Doxis takes over content and archiving from the mainframe and migrates your documents during live operations, so your teams keep working and your mainframe stops paying for storage it doesn't need.
The platform does more with the content once it arrives. AI search helps your teams find the right document faster, and process automation runs approvals and claims workflows on the same content. You can run Doxis in the cloud or on premises, or combine both to match your compliance requirements.
- Lower mainframe costs by moving the content that drives capacity use
- Migration during live operations, with metadata moved first and content following in the background
- A traceable record of every migration step for your auditors
- Retention management and an audit trail for regulated records
- Fast retrieval for customer service, claims and lending teams
- A content foundation ready for AI search and automated workflows
Doxis is a Leader in the Gartner® Magic Quadrant™ for Document Management 2026. Request a free demo to see how Doxis handles your mainframe content.
Automate Work. Accelerate Business.
Bring together AI, ECM, and workflow automation in one powerful enterprise platform.
FAQs on mainframe modernization
Are mainframes becoming obsolete?
No. Mainframes still run core transaction processing at many of the largest banks and insurers, and Kyndryl's 2025 survey found fewer applications moving off the mainframe than the year before.
Can AI replace the mainframe?
AI doesn't replace the mainframe's transaction processing. It supports modernization by helping teams analyze legacy code and by making content searchable once it moves to a modern platform.
Is IBM still making mainframes?
Yes. IBM introduced its z17 mainframe in 2025, designed for AI workloads alongside transaction processing.
What's the difference between mainframe modernization and mainframe migration?
Mainframe modernization is the overall program of updating mainframe systems. Mainframe migration is one part of it: moving selected workloads, such as documents and archives, to other platforms.
What does mainframe to cloud migration involve?
It means moving selected workloads from the mainframe to cloud infrastructure in phases. For documents, that includes moving metadata, migrating content in the background and keeping a full record of every step.
What ROI does mainframe modernization deliver?
Respondents to Kyndryl's 2025 survey reported returns of 288% to 362%, depending on the path they chose.
Which mainframe workloads should move first?
Documents and archives are the strongest first candidates. They drive storage costs, and moving them leaves core transaction logic untouched.
Marc Volquardsen
I am a Product Manager & Solution Architect and have been with Doxis since 2004. After 15 years as a Solution Consultant for Sales, in 2020 I switched to Product Management, where I design solutions for customers based on Doxis, SAP and Salesforce. Please feel free to contact me to talk about solutions for you!
How can we help you?
+49 (0) 30 498582-0Your message has reached us!
We appreciate your interest and will get back to you shortly.