P R I M A R Y C A T E G O R Y : D.2.5 Monitors. C R O S S R E F E R E N C E S : D.4.7 Real-Time Systems, D.4.7 Distributed Systems. Author(s): Hideyuki Tokuda, Makato Kotera, Clifford E. Mercer (Carnegie-Mellon University) "A Real-Time Moni¢or for a Distributed Real-Time Operating System" Title Journal : SIGPLAN N o t i c e s 2 4 , 1 ( J a n . 1 9 8 9 ) , 6 8 - 7 7 . [ S p e c i a l I s s u e : P r o c e e d i n g s o f t h e 1989 SIGPLAN and SIGOPS workshop on parallel and distributed debugging, May 1988] Distributed real-time systems are difficult to develop. External events occur independently of internal control, and the real-time system must be designed to accommodate them correctly. Two problems emerging from this are the logical correctness and the timing correctness of the system software: not only must it process the real-time events correctly, but the program timing must prevent the task of processing from interfering with the task of monitoring. The authors are developing the A R T S distributed real-time operating system; this paper describes some interesting aspects of A R T S : a graphical display for monitoring the behavior of system processes (not particularly innovative in itself), programming language extensions expressing realtime constraints, and real-time debugging capability. The language extensions have an object-oriented flavor, stressing the notion of time encapsulation similar to ordinary data encapsulation; they treat the issue of real-time constraints much better than, say, ADA. A fundamental problem with real-time debugging is invasiveness - collecting debugging information at runtime will disturb the time sequencing and thus change the behavior of the program. Their solution is to monitor the system at all times, regardless of whether the system is actually being debugged, so real-time behavior is preserved. The monitoring events themselves are managed by the real-time scheduler, which prevents interference with system activities. Both these ideas are quite clever; their approach to non-invasive debugging is original. The paper contains a few typos and misnamed references, but the ideas make it more than adequate for the SIGPLAN/SIGOPS proceedings. I would recommend the interested reader contact the authors for papers providing more complete treatment. -- Carl G. Ponder, Livermore CA. Positions Wanted or Available, PER welcomes notices of positions wanted or positions available, and will run them at no charge, space permitting. Please consider the publication schedule for the next issue when sending submissions. February 7 is the submission deadline, after which there is an eight week delay for production. That is, In by February 7, out about the second week in April (in the member's hands). Performance Evaluation Review Vol. 18 #3, November 1990
/lp/association-for-computing-machinery/a-real-time-monitor-for-a-distributed-real-time-operating-system-AroZfk7FVe