Enterprises are challenged with dependencies on legacy mainframe systems. These legacy solutions pose many business risks including;
- difficulty in maintaining and operating these legacy systems as they were built in early 1980s with years of applied updates and patches,
- the ability to add functions because subject matter experts for these systems are rapidly retiring,
- significant increase in data volume, data rate, and data types is challenging these systems, and
- difficulty applying new data analytics as data is locked in legacy data stores.
We are observing that the transition from legacy platforms to Hadoop based platform is much more feasible compared to prior efforts done using client server, middleware, or web based technologies. The reason for this migration success is because of many similarities between Hadoop technologies and legacy systems. Some of the similar characteristics are following;
- Like legacy applications, mature part of Hadoop is based on batch processing (Map-reduce jobs).
- Data in legacy applications is kept separate from data schemas for e.g. in COBOL data sections captured as COBOL Copybooks are seperate from programs. Hadoop uses a similar data seperation approach but applies data schemes at read time in MapReduce or Hive datastore.
- Like Mainframes, Hadoop is a single system even though it utilizes underneath a distributed storage and parallel processing architecture.
Because of these similarities we can significantly automate migration of these applications from legacy to Hadoop. This migration hence realizes big data benefits such as;
- Enable quicker data processing and hence reduce time to market data for analytics,
- Enable system owners to address data volume, variety, and data rate issues
- Significantly reduce and manage cost in storing and processing,
- Enable new analytics algorithms such as machine learning on full data set utilizing multimode Hadoop architecture
- Get business commitment with quick value demonstrations without significant cost or risk, and
- Utilize modern tools to rapidly ingest and map data for processing using existing data logic and data schema such as COBOL code/JCL and COBOL copybooks.
- Description of legacy system
- Legacy system’s risks and challenges
- Hadoop similarities to Mainframe
- Migration effort
- Sample customer usage scenarios
- Realized benefits