Google文件系统（国外英文资料）.docVIP

Google文件系统 gfs GFS is an extensible, distributed file system for large, distributed, and large data access applications. It runs on inexpensive generic hardware, but provides fault tolerance. It can provide high performance services to a large number of users. 1. Design Overview (1) design scenario GFS has many of the same goals and distributed file systems of the past, but the GFS design is driven by the workload and technical environment of current and anticipated the application, which reflects the idea of it is obviously different from the file system in the early. This requires the re examination of traditional choices and the exploration of completely different design perspectives. The different views of GFS and previous file systems are as follows: 1, component errors are no longer treated as exceptions, but are treated as common cases. Because the file system is made up of hundreds of machines for storage, these machines are made up of cheap common components and accessed by a large number of clients. The number and quality of components make it possible for some machines to work at any time and some may not be able to recover. Therefore, real-time monitoring, error detection, fault tolerance and automatic recovery are essential to the system. 2, according to the traditional standards, the documents are very large. Files of several GB lengths are quite common. Each file usually contains many application objects. It is difficult to manage thousands of KB sized file blocks, even if the underlying filesystem supports, when dealing with rapidly growing datasets that contain tens of thousands of objects and TB in length. Therefore, the parameters and size of the operations in the design must be reconsidered. The management of large files must be efficient, and small files must be supported, but not optimized. 3. Most file updates are done by adding new data rather than changing existing data. In a file, random operations practically do not exist in practice. Once