voula@utcsrgv.UUCP (Voula Vanneli) (02/11/85)
UNIVERSITY OF TORONTO DEPARTMENT OF COMPUTER SCIENCE (SF = Sandford Fleming Building, 10 King's College Road) (MC = McLennan Physical Laboratories, 60 St. George Street) (RS = Rosebrugh Building, Taddlecreek Road) SYSTEMS SEMINAR - Tuesday, February 19, 10 a.m., SF 1105 Dr. Jim Gray Tandem Computers, Cupertino, CA "Approaches to Fault Tolerant Computing" * ABSTRACT First an overview of the relative importance of human, software and hardware faults is presented. I argue that reliable hardware is a reality and that most system failures are due to software and human errors. Simple user interfaces are the solution to human errors. Several approaches to tolerating software errors are presented; transactions, lock-step process pairs, shadow process pairs, and persistent process pairs. It is argued that software errors are soft and hence transactions plus persistent process pairs are the best solution. Discussion then turns from fault tolerant execution to fault tolerant storage. It is shown why data mirroring is more relevant than exact replicas and why majority replica schemes are not appropriate in a local or long-haul network. I argue that schemes for long-haul replication require inconsistency and then present the Snap Shot and ASAP replica approaches with examples.
wlrush@water.waterloo.edu (Wenchantress Wench Wendall) (05/11/89)
DEPARTMENT OF COMPUTER SCIENCE UNIVERSITY OF WATERLOO SEMINAR ACTIVITIES SYSTEMS SEMINAR -Friday, May 12, 1989 Motoshi Saeki, Tokyo Institute Technology, Faculty of Engineering, International Institute, Fujitsu Limited, will speak on "Software Development Process from Natural Language Specification". TIME: 1:30 p.m. ROOM: DC 1304 ABSTRACT A process to derive incrementally a formal specification from an informal specification written in natural language is presented. This process consists of two major activities, "design" and "elaborate". Through the design activity, the structure of software modules based on an object oriented model is interactively extracted from the informal English description. Each word such as nouns and verbs in the natural-language sentences is associated with a software concept, e.g. class, attribute, and method. Especially we concentrate on types of verb patterns occurring in the sentences to increase applicability to dynamic system. The elaborate activity is one to refine and rewrite the informal specification more precisely based on the derived module design document. We can complete the formal specification through the elaborate - design activity cycle. Each document created in our process is called product. The integration management of the various kinds of the products are supported by a hypertext system. Furthermore we discuss the whole of software development process in which the design process and the elaborate one are embedded.
wlrush@water.waterloo.edu (Wenchantress Wench Wendall) (11/17/89)
DEPARTMENT OF COMPUTER SCIENCE UNIVERSITY OF WATERLOO SEMINAR ACTIVITIES SYSTEMS SEMINAR -Thursday, November 23, 1989 Dr. Mohammad Malkawi, Dept. of Electrical Engineering and Computer Science, University of Wisconsin-Milwaukee will speak on ``Load Balancing in Distributed Systems.'' TIME: 3:30 p.m. ROOM: DC 1304 ABSTRACT In a distributed network, it is very important to balance the load among the processors in the network such that the average response time in the system is imroved. A new load balancing algorithm has been developed based on graph coloring. Each link connecting any two nodes is assigned a color such that the links adjacent to any one node are colored with different colors. To balance the load of the system, adjacent processors connected with a link of color C share i the loads between them such that their load difference is reduced to a value <^H_ 1. To balance the load in the system, all colors have to be considered in a cyclic order. The process of balancing continues until the difference between the loads of adjacent processors is no more than 1. The described algorithm is distributed, dynamic, and requires no global knowledge of any type. Simulation results will be discussed. The algorithm achieves significant improvements in response time and throughput. Simulation is performed for both static and dynamic systems. Different networking topologies are considered including random graphs and n-cube structures. November 15, 1989
wlrush@water.waterloo.edu (Wenchantress Wench Wendall) (02/27/90)
DEPARTMENT OF COMPUTER SCIENCE UNIVERSITY OF WATERLOO SEMINAR ACTIVITIES SYSTEMS SEMINAR -Friday, March 2, 1990 Professor Jay Black, Dept. of Computer Science will speak on ``Abstraction for Debugging Distributed Programs.'' (This is a joint work with W.H. Cheung). TIME: 10:30 a.m. ROOM: DC 1331 ABSTRACT Distributed programs are evolving towards greater size and complexity. A non-trivial distributed program consists of a large number of cooperating dynamic processes. Not only do distributed programs tend to involve more code than their sequential equivalents, but they also suffer from further increases in complexity due to the presence of concurrency, non- determinism, and the possibility of partial failure of processes, nodes, and communication channels. Debugging distributed programs is thus a difficult endeavor, and new techniques are required to manage this more complex environment. One of the most powerful tools we know for controlling complexity is abstraction, and we provide a rigorous basis for the use of it in distributed debugging, and hence in distributed systems in general. We will discuss how abstraction can be applied both to processes (forming "clusters" or process groups), and to events occurring in the system. This talk will serve as a starting point for a discussion on current and possible future directions for the Shoshin distributed debugger prototype design. Any interested people are welcome.