The proxy design pattern revisited
Pattern languages of program design 2
Effective distribution of object-oriented applications
Communications of the ACM
The Rational Unified Process: an introduction
The Rational Unified Process: an introduction
Client/Server Performance Tuning: Designing for Speed
Client/Server Performance Tuning: Designing for Speed
A performance tuning framework
ISTA '01 Proceedings of the 2001 international conference on Information systems technology and its applications - Volume P-2
| Hi-index | 0.00 |
Despite advances in hardware, communications and software technology, latency remains a pressing problem for designers of distributed applications. Even after application functionality has been carefully distributed amongst multiple sites, there is the problem of the detailed design of client requests so that network trips are efficiently made. Most solution approaches (or compromises) for detailed design that are encountered in practice either tend to be ad-hoc, or are intended for use with a specific class of development environment, such as object-oriented languages. In this paper we take a fresh look at the problem of designing client requests in the presence of latency. We treat the client machine as a black box, thus ignoring the details of intra-machine communication and staying clear of specific environments such as OO. We propose a generic, intuitively appealing principle for the sequential design of client requests that allows a designer to systematically work around latency to meet response-time requirements. Applicability of this principle is shown using some simple but realistic examples of business transactions.