The local DAMA chapter in Dublin, OH, invited me to present Virtualization and Big Data to the group. I'm set up in the Lakeside Conference room at the OCLC conference center.
Early in my career, I was a software engineer for DataPhase systems. One of my tasks was designing and implementing a MARC record database for the company's Automated Library Information System (ALIS). The system was implemented in MIIS, Meditech's implementation of the MUMPS system. MUMPS is now known as simply "M" and MARC is the U.S. Library of Congress machine readable cataloging record format.
While I'm waiting for the conference attendees to appear, I'm thinking about how difficult it was to deal with the MARC record format. A MARC record is made up of a variable number of tags containing a certain type of data. Tags can either be in a record or not. They can appear multiple times. This allows the record format to deal with such things as a book or periodical having multiple editors, authors, or publishers. It also makes it possible for a document to have multiple titles and subtitles.
MIIS, as well as all other MUMPS implementations, is based upon a sparse array database. Records can be text or numbers and can be reset at any time simply by storing something new there. Furthermore, a record can contain data, be a pointer to a lower level in the database or both. As with the record format, this can easily be changed by storing something else.
It was wonderful to revisit my time as a MUMPS software engineer and my labors to write code that read an OCLC database tape or scrape an OCLC terminal display for database information.