First, the inode and directory entries
The same CPU, the same memory, disk and file system management, as well as the core operating system features.
(1) disk to provide the most basic of persistent storage
(2) On the basis of the file system on the disk, providing a tree structure for managing files, then the disk and the file system is how it works? And what indicators can measure their performance?
We all know that everything is a file in Linux, not only ordinary files and directories, and even block device, socket, and management will all have to be managed through a unified file system.
To facilitate the management, Linux file system for each file is assigned two data structures: inode (index node) and directory entry (directory entry).
（3）
Inode : iNode for short, is used to record the metadata files, such as iNode number, file size, access, modification date, location data and so on.
Inode and file correspondence, file it with the memory, as will be persistent storage to disk, so the disk space occupied by the same inode.
Directory entry : short relationship dentry, inode pointer as well as with other directory entries.
In other words: inode is the only flag for each file and directory tree structure is the key to maintain the file system. The relationship between directory entries and inode is many, that is: a file can have multiple aliases.
First: directory entry itself is a memory cache, inode data is stored on disk.
The inode will naturally be cached in memory, speed up file access