| With the development of big data, the requirements of storage system are getting higher. Multitask sharing rise the system concurrency and data requests showed fragmentation characteristics. Solid State Driver has received extensive attention due to its excellent reading and writing performance. However, the asymmetry of its read and write performance and cost consideration make the hybrid storage system of SSD-HDD the main direction of research.By analyzing the characteristics of SSD and HDD, we presents a scheduling algorithm based on dynamic replacement costs(DRC), fully considering the hot data in the request and the cost of replacing dirty data, not only effectively improve the cache hit rate, and reduce random write operations in HDD in order to improve the overall performance of the system.This paper focuses on the following aspects:Three different kinds of storage device are studied: Hard disks, Solid State Disk, NVRAM. We introduced their different I/O performance and their limitations.Key technology of using Solid State Disk is studied, and presents other cache management algorithms that are now applying in the hybrid storage systems.Show the detail of a hybrid storage system which is based on Device Mapper, and analyze the read and write scheduling mechanism, the underlying implementation mode and the main metadata structure.Design and implementation of the DRC algorithm, so that it can better cope with the cloud computing scenarios, and optimized the complexity of DRC algorithm.Production workloads are used to test and compare the different I/O performance among FIFO, LRU and DRC, the experimental data shows that DRC works well in the hybrid storage system, and achieve the expected goals.The experiment results show that in high concurrency scene, DRC algorithm enhance the hit rate up to 11.6%, and I/O speed up 16.7% compare to LRU or FIFO. In a variety of cache size conditions DRC has achieved remarkably enhance performance... |
PDF Full Text Request