Internet of things (IoT) consisting of sensors including smart meters generates and transports massive data at high velocity. Distributed data stream processing system can parallelize the procedure of data processing, and improve the efficiency of data analysis. However, existed data stream processing systems are not specially designed for contexts with high workload like smart grid. When workload become very high, the performance of the system will decrease rapidly. Toward this issue, an IoT-oriented distributed stream processing framed is proposed. The frame is designed on the basis of distributed event stream system, which is applicable for real-time big data in smart grid. A hierarchical architecture is exploited, making the proposed frame more scalable. Experimental results show the low latency and high throughput of the designed system.