Technical reports

Monitoring composite events in distributed systems

Scarlet Schwiderski, Andrew Herbert, Ken Moody

February 1996, 20 pages

DOI: 10.48456/tr-387

Abstract

One way of integrating heterogeneous, autonomous and distributed systems is to monitor their behaviour in terms of global compostite events. In specific applications, for example database, it is essential that global composite events can take account of general conditions such as the timing constraints on distributed system behaviour. In this paper the use of global composite events incorporating time events for expressing physical time is investigated. The detection of global composite events is complicated by the inherent features of distributed systems: lack of global time, message delays between sites and independent failures. Global event detectors are distributed to arbitrary sites. Relevant constituent events occur on remote sites and are signalled to corresponding global event detectors, where they are evaluated. Two different algorithms for the detection of global composite events are introduced which are based on the evaluation of trees: asynchronous and synchronous evaluation. Asynchronous evaluation provides fast but unreliable detection of global composite events, whereas synchronous evaluation is characterized by reliability and unpredictable delays.

Full text

PDF (1.7 MB)

BibTeX record

@TechReport{UCAM-CL-TR-387,
  author =	 {Schwiderski, Scarlet and Herbert, Andrew and Moody, Ken},
  title = 	 {{Monitoring composite events in distributed systems}},
  year = 	 1996,
  month = 	 feb,
  url = 	 {https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-387.pdf},
  institution =  {University of Cambridge, Computer Laboratory},
  doi = 	 {10.48456/tr-387},
  number = 	 {UCAM-CL-TR-387}
}