My relationship with Multics was quite varied. I was a graduate student working for Corby in Tech Square, had a summer job with CISL and then a full-time job with HIS both across the Tech Square plaza. During that time I worked on the following:
- Paging performance on disk and drum. Published a paper on performance optimization of the "drum folding" algorithm in the 6th Sigops Symposium in 1973. Performed lots of late-night experiments on paging rates and alternate channel configurations, leading to my MS/EE thesis "Performance Evaluation of Rotating Disks in a Multiplexed Computing Environment"
- Wrote a subject evaluation survey processing application for the MIT EE department sponsored by Eta Kappa Nu honor society.
- Wrote major parts of Prof. Roy Kaplow's computerized learning system project (TICS).
- Wrote significant enhancements to the core syserr log system to support secure system auditing for military use of the system.
After Multics, I joined Max Smith at Prime Computer, which we used to call "Multics in a Matchbox". Got to reinvent some wheels in a minicomputer form. There I worked on networking architectures, office automation, workstation architectures, advanced development of modular computing architectures. While at Prime I helped to found the Object Management Group (OMG). I take credit for drafting the first requirements sketch for what came to be called an Object Request Broker (ORB). At Prime I worked also with another ex-Multician, Barry Wolman.
From Prime I spun off a startup company, Constellation Software, to productize remote database access middleware based on modular component oriented computing ideas. Unfortunately, we were underfunded and our lunch was eaten by the much-less-capable-but-free ODBC. Ah, well! We sold our startup to VMark, which produced a Unix version of the PICK operating system and database.
At VMark I became the initial and chief architect of an internal startup called DataStage. Originally this technology applied modular computing and visual programming ideas to the extraction, transformation and loading (ETL) problem of building analytic data warehouses from heterogeneous (and often grubby) operational data sources. Through numerous mergers, acquisitions, divestitures and more acquisitions (VMark->Ardent->Informix->Ascential), this product has evolved and grown into the industry's premier data integration product suite, and I have emerged as Chief Technology Officer and chief architect.
Through all this, I echo some of Corby's Multics "lights out" letter comments - after 35 years, the industry is still learning lessons first explored in Multics.