Sliding window technology for data streams is widely used to finish the data stream sampling. In this paper, rewritable circular sliding window, ring circular sliding window and sliding window based vector model are introduced. Experimental results show that rewritable circular sliding window is faster than sliding window based on vector model. But the speed advantage is slackening up, with the increasing of sliding window. Data stream sampling based on ring circular sliding window is the fastest. With the change of sliding window, the speed is smooth. Experimental results show that the data stream sampling model based on ring circular sliding window is proper for the migration and mining of data stream.