page 1  (15 pages)
2to next section

Quality of Service and Scientific Workflows*

Mladen A. Vouk
Software Engineering and Multimedia Laboratories
Department of Computer Science, Box 8206
North Carolina State University, Raleigh, NC 27695
Tel: 919-515-7886, FAX: 919-515-6497, e-mail: vouk@adm.csc.ncsu.edu

and

Munindar P. Singh
Database Laboratory
Department of Computer Science, Box 8206
North Carolina State University, Raleigh, NC 27695
Tel: 919-515-5677, FAX: 919-515-7896, e-mail: singh@adm.csc.ncsu.edu

(To be presented at the IFIP/TC2/WG 2.5 Working Conference 7 - "Quality of Numerical Software: Assessment and Enhancements," Oxford, U.K., 8-12 July 1996. Revised version will appear in the WoCo7 proceedings book that will be published by Chapman & Hall 1996).

Key Words
quality of service, QoS, problem solving environment, PSE, scientific workflow, reliability, availability, intra- and inter-net response delays, users.

Abstract

This paper discusses some quality issues that face the next generation of the network-based scientific problem-solving environments. The advent of high-performance computing engines and networks is opening a fantastic opportunity for bringing serious numerical and problem-solving applications closer to a broad base of potential users with widely differing needs. From the perspective of these users, a key issue will be the quality of service (QoS) in the broader sense, i.e., not just parameters such as network delays and throughput, but also end-user quality factors such as system availability, system functionality and content quality, and so on. In order to facilitate integration of the QoS and PSE concepts we view PSEs through the prism of scientific workflows, i.e., a series of structured activities and computations that arise in scientific problemsolving. Scientific workflows are expected to coexist and cooperate with other user workflows (e.g., business workflows, educational workflows, legislative workflows). As such they must support compatible QoS. We are using data from existing network-based systems and workflows to quantitatively bound some of the PSE QoS parameters. For example, system availability must be 0.95 or better, while adequate synchronous user-machine interactions require user-level end-to-end (round-trip) delay times that are consistently less than about 250 msec, and network-level roundtrip delays that do not exceed 100 to 150 msec. Use of multimedia imposes additional restrictions, while end-user risks impose bounds on the security and reliability of numerical computations and algorithms. It is our belief that the next generation of PSEs must have QoS parameters designed into the system, or these PSEs will fail to live up to user needs and expectations.

* Research supported in part by NSF (ACS 9418960 and ACS 9696131), NASA (NAG-1-983), EPA/MCNC (94- 7050-027/5-30012 and 94-7050-029/5-30816), IBM Canada (CAS) Fellowship, IBM Corp. (RTP), and CACC (61- 0103/5-30456 and 83-0062/5-30781).