Absrtact: MapReduce technology for large-scale data analysis has been widely recognized and researched, we have seen the explosive growth of cloud data service system. These new systems are so-called "cloud OLTP" applications, although they do not support acid transactions. Typical examples such as http://www.aliyun.com/zixun/aggregation/14239.html ">bigtable,pnuts,cassandra,hbase,azure,couchdb, Simpledb,voldemort, wait. And they are applied to many areas (such as TPC) that are different from traditional service workloads. The number of emerging cloud serving Bae and the wide range of proposed applications,coupled with a lack of apples-to-apples Configured Comparisons,makes it difficult to understand the tradeoffs inclusive the BAE and the workloads for abound tightly Suited. (Emerging cloud systems with a wide range of other systems, because of the lack of fair performance comparisons, to judge among systems who are more suited to their workloads) we released the Yahoo! The Cloud serving Benchmark (YCSB) framework is designed to facilitate performance comparisons of new cloud data service systems. We have defined a core set of benchmarks and reports for four widely used systems: The Cassandra,hbase,yahoo! ' s pnuts and the simple implementation version of MySQL. Through open source, we hope that other