Dynamic Load Balancing of Virtualized Database Services Using Hints and Load Forecasting

Abstract. Future database application systems will be designed as Service Oriented Architectures (SOAs), in contrast to today’s monolithic architectures. The decomposition in many small services allows the usage of hardware clusters and a flexible service-to-server allocation but also increases the complexity of administration. Thus, new administration techniques like our self-organizing infrastructure are necessary. It monitors the system, reacts automatically on exceptional situations, e.g., overload of a server, and features self-optimizing capabilities. In the context of distributed services it takes some time until the reaction on an exceptional situation becomes effective. During this time the server stays overloaded which leads to a poor performance of services running on this server. In this paper, we present a novel concept to improve self-organizing infrastructures to react proactively. For this purpose we present two techniques: Short-term load forecasting for services w...

• Type: Conference
• Year: 2005
• Where: CAISE
• Authors: Daniel Gmach, Stefan Krompass, Stefan Seltzsam, Martin Wimmer, Alfons Kemper
• Keywords: CAISE 2005, Exceptional Situation, Overload Situations, Self-organizing Infrastructures