San Raid Controller-based Design And Realization

With the rapid development of computer and network technology, computers are used more and more extensively and deep in various industries. In most enterprises and the majority of industries, there are some critical applications, which must be 7d*24h uninterrupted running. Under such circumstances, the system's high availability is particularly important, and the higher requirements for storage technology are put forward.SAN technology has advantage of its own features and is used more and more widely. The data is stored in storage devices in the SAN network. At present, the storage device always is Redundant Arrays of Independent Disks, which can increase the reliability and security of data greatly. At the same time, there is still some space to upgrade in fault-tolerant capabilities.1. Although, the development of hardware technology greatly enhanced the reliability of the system, the failure of the other core components of the system (such as CPU, motherboard, physical memory, etc.) may occur, when applications will have downtime of 44 to 87 hours in 365 days a year. This requires considering raising the high-availability of system in a higher level and from more aspects.2. When system crashes, power outages or Bad Track occurs in disks, the data in RAID may be destroyed and the stripe data is inconsistent. And then, if some accidents happened in a disk, and lead the disk to fault, the RAID controller will not be able to get the correct data from calculate parity redundant data, and can not ensure the consistency of data after reconstruct also. At worst, it is of seriously consequences that data errors or data lost are resulted in.Based on the deeply study of the variety key technologies of RAID, the solution of dual controller active-active and stripe verify and repair strategy were carried out. Through running plenty of function test cases, it is proved that the RAID controller can adapt to environmental changes smoothly and the data is available, regardless in the normal mode or in the abnormal mode. And, under normal circumstances, the design of dual control can also share the load and improve efficiency. When a controller is abnormal, another controller can quickly take over the business. In this process, the efficiency of I/O will be decreased, but the duration of the process is not too long. So, the original intention of the design is achieved. The stripe verify strategy can find inconsistencies in the stripe data level, meanwhile, The stripe verify repair strategy is able to repair the stripe data inconsistency, which can prevent the spread of the error, and improve the fault tolerance of RAID controller.