On-demand computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources. It provides several services. On-demand computing based transaction processing is one of them which deals with the challenge to an enterprise to meet fluctuating demands of sufficient resources efficiently. Concepts such as grid computing, utility computing, autonomic computing, cloud computing and adaptive management seem very similar to the concept of on-demand computing. On-demand computing, which is a form of distributed computing involves coordination and sharing of computing resources across the web globally. On-demand computing based transaction is a group of operations executed in on-demand computing platform to perform some specific functions by accessing and updating a database. The overarching goal of this dissertation is to propose soft computing based load balanced scheduling and allocation techniques which can make on-demand computing based transaction processing system dependable.