The Software Modeling and Implementation of Reliable Server Pooling and RSPLIB

With the growing complexity of software applications, there is an increasing demand for solutions to distribute workload into server pools. Grid Computing provides powerful - but also highly complex - mechanisms to realize such tasks. Also, there is a steadily growing number of downtime-critical applications, requiring redundant servers to ensure service availability in case of component failures. To cope with the demand for server redundancy and service availability, the IETF has recently standardized the lightweight Reliable Server Pooling (RSerPool) framework, which is a common architecture for server pool and session management. In this paper, we first introduce the concept of RSerPool and then present the modeling thoughts of RSPLIB and the underlying general groupware design. Based on RSPLIB, we will illustratively show how to easily develop applications on top of RSerPool. We will also offer an application evaluation example for a proof-of-concept setup to distribute ray-tracing computation workload into a compute pool.