A study of throughput degradation following single node failure ina data sharing system

摘要

The data sharing approach to building distributed database systemsis becoming more common because of its potentially higher processingpower and flexibility compared to data partitioning. However, due to thelarge amounts of hardware and complex software involved, the likelihoodof a single node failure in the system increases. Following a singlenode failure, some processing has to be done to determine the set oflocks held by transactions which were executing at the failed node.These locks cannot be released until database recovery has completed onthe failed node. This phenomenon can cause throughput degradation evenif the processing power on the surviving nodes is adequate to handle allincoming transactions. This paper studies the throughput dropoffbehavior following a single node failure in a data sharing systemthrough simulations and analytical modeling. The analytical modelreveals several important factors affecting post-failure behavior and isshown to match simulations quite accurately. The effect of hot locks(locks which are frequently accessed) on post-failure behavior isobserved. Simulations are performed to observe system behavior after theset of locks held by transactions on the failed node has been determinedand show that if the delay in obtaining this information is too large,the system is prone to thrashing