Fujitsu Laboratories develops technology to automatically resolve performance problems in a distributed storage system
Fujitsu Laboratories today announced the development of a technology that automatically resolves problems caused by concentrated accessing of popular data items in a distributed storage system in order to curb access-time slowdowns. Distributed storage combines multiple servers into a single storage. Increasing the number of servers improves storage capacity and performance, making this approach appropriate for storing data that grows day by day. Furthermore, storing replicas of the same data simultaneously on multiple servers increases data reliability and access performance. If, however, there is a sharp increase in accesses to a particular stored data item, the load on the server storing it will increase, which may greatly cause an upturn in user access times.
Fujitsu Laboratories has developed a technology that can instantly detect spikes in popularity for a data item and automatically increase the number of replicas of it to level-off server loads. This automates what had previously been a manual approach for dealing with popularity spikes, and can limit slowdowns in access times. The new technology applies to distributed object storage , and in test cases of access concentrations on the Internet, has been proven to ease those concentrations by approximately 70%, resulting in an improvement in access times of tenfold or greater. This technology makes it possible to stabilize operations on ICT systems where access patterns are difficult to predict.