| Today,the fast-growing Internet of Things(IoT)applications need responsive and efficient data processing and more power-hungry warehouse-scale computers(WSC)in the next decade.Traditional cloud datacenters will face a bulk data aggregation challenge,which not only hinders timely data analytics but also leads to networking bandwidth issues.Building even larger compute cluster near the data may temporarily alleviate the problem,but the data will quickly saturate these edge machines.Due to insufficient capacity,they still have to heavily rely on the cloud and spend a significant amount of time and energy on data movement.In this work we focus on the cause of inefficiency in IoT big data management and propose an optimized architecture and energy saving method for IoT datacenters.To maximize the operation efficiency,we have designed an internal controller for energy savings.It will change the operating states of datacenters to optimize the overall efficiency and responsiveness.In addition,we apply a combination of software and hardware method for datacenter optimization.Our system prototype shows that our approach is able to maintain system performance by 80% compared to conventional cloud datacenter,while saving 50% energy. |
PDF Full Text Request