A software lease is a temporary contract between the client (lessee) and the server (lessor) that binds the server to make resources accessible to the client and provide certain services (such as maintaining the object's consistency) during the time of a lease. Virtually every Web-based transaction (think eBay, Amazon.com, Expedia) and mobile applications (Lotus Notes) involves use of leases.
However, in order to use leases effectively, a program designer is required to estimate the required lease duration. Short leases permit the resource to recover quickly in the case of the client failure, but increase the message load for both clients and servers. Long leases require less renewal traffic, but suffer the same responsiveness problems as locks. This work aims at quantifying the above tradeoff and analyzing the gain in performance when lease duration is assessed correctly.
Two significants steps have been already made: a) creation of an analytical model for leases and b) development of a prediction system for the case when the lessee functionality is encapsulated in an OO component and available for analysis. These results as well as preliminary evaluation of the model are described in the following research report.
The student would need to study the problem and understand the technology. Then, the next important task would to to devise and run the experiments that should reflect on the model. In addition, there is greater research potential for an interested student, e.g., improving the prediction technique, refining the model, considering other lease applications for evaluation, etc.
Program
Informatikk