Building Enterprise Class Real-Time Energy Efficient Decision Support Systems

In today’s highly competitive marketplace, companies have an insatiable need for up-to-the-second information about their business’ operational state, while generating Terabytes of data per day [2]. The ability to convert this data into meaningful business information in a timely, cost effective manner is critical to their competitiveness. For many, it is no longer acceptable to move operational data into specialized analytical tools because of the delay this additional step would take. In certain cases they prefer to directly run queries on their operational data. To keep the response time of these queries low while data volume increases, IT departments are forced to buy faster processors or increase the number of processors per system. At the same time they need to scale the I/O subsystem to keep their systems balanced. While processor performance has been doubling every two years in accordance with Moore’s Law, I/O performance is lagging far behind. As a consequence, storage subsystems not only have to cope with the increase in data capacity, but, foremost, with the increase in I/O throughput demand, which is often limited by the disk drive performance and the wire bandwidth between the server and storage.