Chapter 10. The Legacy Zone

In the late 1990s, I was working for a company as a part of its e-commerce development team. Our job was to build an Internet front end to the data that was generated by the company's COBOL-based legacy systems. After various options were explored, the final solution that was reached between us (the Internet team) and them (the mainframe guys) was that they would export all their data into a SQL server database. Then we would be free to display it however we chose, using normal Web-development techniques.

After the project was launched, I became curious about how they planned to import the data into the SQL database. I met with the tech lead of the back-end system, and he showed me what they were doing. Their solution was to develop a number of brand new COBOL applications that would run on an NT server that would import flat files into the SQL database. When I asked them why they chose COBOL, the answer I received was, “It's the only language my team knows.” This drove home the message that the death of the mainframe and the mainframe programmer has been greatly exaggerated.

In applications such as banking, stock trading, and utility billing, key transactions are still served by mainframe applications. Despite the high cost of ownership associated with mainframe installations, the immense investment in custom applications and the extremely high reliability of mainframe hardware makes it difficult for large IT departments to justify migrating to a more modern server architecture.

This chapter is written from the point of view of a non-mainframe programmer who needs to interface an application with data stored in a legacy environment. Similar to the old saying about pilots who land with their landing gear retracted, “There are those who have, and those who will.” But giving some thought to the proper tools and protocols in advance can make the process much less painful for all concerned.