The Design of Multimedia Object Support in DEC Rdb

1 Abstract Storing multimedia objects in a relational database offers advantages over file system storage. Digital's relational database software product DEC Rdb supports the storing and indexing of multimedia objects-text, still frame images, compound documents, audio, video, and any binary large object. After evaluating the existing DEC Rdb version 3.1 for its ability to insert, fetch, and process multimedia data, software designers decided to modify many parts of Rdb and to use write-once optical disks configured in standalone drive or jukebox configurations. Enhancements were made to the buffer manager and page allocation algorithms, thus reducing wasted disk space. Performance and capacity field tests indicate that DEC Rdb can sustain a 200-kilobyte-per-second SQL fetch throughput and a 57.7-kilobyte-per-second SQL/Services fetch throughput, insert and fetch a 2-gigabyte object, and build a 50-gigabyte database. 2 Introduction To accommodate the increasing demand for computer storage and indexing of multimedia objects, Digital supports multimedia objects in its DEC Rdb relational database software product. This paper discusses the improvements over version 3.1 and presents details of the new features and algorithms that were developed for version 4.1 and are used in version 5.1. This advanced technology makes the DEC Rdb commercial database product a precursor of sophisticated database management systems. Multimedia objects, such as large amounts of text, still frame images, compound documents, and digitized audio and video, are becoming standard data types in computer applications. Devices that scan paper, i.e., facsimile machines, are inexpensive and ubiquitous. Devices that capture and play back full-motion video and audio are just beginning to reach the mass market. Capturing these objects for use within a computer results in many large data files. For example, one minute of digitized and compressed standard TV-quality video requires approximately 50 megabytes (MB) of storage! To date, relational databases have been used successfully in storing, indexing, and retrieving coded numbers and characters. Relational algebra is an effective tool for reorganizing queries to reduce the number of records, e.g., from 1 million to 70 records, that an application program must search to obtain the desired information.