Flow based dynamic time-out

Abstract

Described herein are a system and techniques for increasing efficiency of communications and quality of a user's experience by implementing dynamic time-out intervals. When a service slows due to demand, malfunction on the server side, or local constraints near the client such as local wireless network connectivity, a client-side time-out interval can be dynamically increased so that clients are less likely to depart before receiving requested services. A server that cannot provide requested services can indicate that a client-side time-out interval should be zero so that clients do not wait on a request that will not be fulfilled. A server seeing service requests but few completions or satisfied service requests can indicate different time-out intervals to different clients, observe which intervals have higher completion success rates and then indicate revised time-out intervals accordingly.