Mainframe and legacy data, streaming to your lakehouse
Your most valuable data still lives on mainframe and legacy systems, but analytics, AI and digital teams need it in a modern lakehouse in near real time. ABMartin designs and builds the change-data-capture and streaming pipelines that get it there continuously, without nightly batch extracts, without rewriting source applications, and with privacy protection built in.
The challenge
- Batch extracts leave analytics hours or days behind the business.
- Legacy formats (packed decimal, EBCDIC, copybooks, hierarchical records) don't map cleanly to modern platforms.
- Every new downstream consumer adds another point-to-point feed to maintain.
- Sensitive data ends up copied into cloud platforms without consistent protection.
What changes
- Changes are captured from source logs as they happen, with minimal load on production.
- One governed event stream feeds many consumers: lakehouse, warehouse, microservices.
- Source systems keep running unchanged while modernization proceeds around them.
- Masking and privacy controls are applied consistently once data lands.
How it works
Change data capture reads inserts, updates and deletes from DB2, VSAM, IMS and other legacy sources, using IBM tooling or the CDC platform you already run.
Events are published to Kafka, whether Confluent or another distribution, with schemas, topics and ordering designed for reliable replay.
Data lands in your lakehouse or cloud warehouse, such as Databricks or Snowflake, transformed from legacy layouts into analytics-ready tables.
Sensitive fields are masked or de-identified with consistent rules, using XDP, IBM Optim or your existing privacy tooling.
From architecture to production
Our team has designed streaming architecture and infrastructure for mainframe data, and brings decades of hands-on experience with legacy data formats and data privacy.
Assess
Inventory sources, data volumes, change rates and downstream use cases, and identify sensitive data that will need protection.
Architect
Choose capture, streaming and landing patterns that fit your existing tools, topology and security requirements.
Build
Configure capture, map legacy record layouts to target schemas, and stand up topics, connectors and landing tables.
Protect
Apply deterministic, format-preserving masking so data stays realistic and joinable across systems.
Validate
Reconcile source and target, test latency and recovery, and verify that no sensitive data leaks downstream.
Enable and support
Hand over runbooks and knowledge transfer, or keep us engaged for ongoing operations and new sources.
Running IBM InfoSphere Data Replication with Confluent? See our IBM mainframe streaming approach on the IBM Data Solutions page.
Ready to feel data confident?
