The development of zero-copy technology has different applicable methods in different scenarios. In the Linux operating system, there are many existing zero-copy technologies, most of which exist in different Linux kernel versions, and some old technologies have been replaced by new technologies. This article will divide the applicable scenarios of these zero-copy technologies, and introduce the three main zero-copy technologies in Linux in detail.
Direct I/O
Direct I/O is a data transmission method, the application program can directly access the hardware storage, and the operating system kernel is only an auxiliary data transmission. This technique is suitable for situations where the operating system kernel does not need to directly process the data. Data can be directly transferred between the buffer in the address space of the application program and the disk, without the support of the page cache provided by the Linux operating system kernel. The main purpose of this technique is to avoid buffer copy operations between the application program address space and the operating system kernel address space.
Avoid copy operations
During data transmission, avoid copying data between the buffer in the operating system kernel address space and the buffer in the user application address space. If the application program does not need to access the data during data transmission, copying the data from the Linux page cache to the buffer of the user process can be completely avoided, and the transmitted data can be processed in the page cache. In some special cases, this zero-copy technology can achieve better performance. The similar system calls provided in Linux mainly include mmap(), sendfile() and splice().
copy-on-write
The copy-on-write technology focuses on flexibly handling the copy operation of data between the buffer of the user process and the page cache of the operating system. This technology optimizes the data transmission process between the Linux page cache and the buffer of the user process, which continues the traditional communication method, but is more flexible. In Linux, this method mainly utilizes copy-on-write technology. The main purpose of the copy-on-write technology is to optimize the efficiency of data transfer between the user address space and the operating system kernel address space.
These three zero-copy technologies are widely used in Linux. The purpose of the first two techniques is mainly to avoid the buffer copy operation between the address space of the application program and the address space of the operating system kernel. These two types of technologies are generally applicable to some special cases, for example, the data to be transmitted does not need to be processed by the operating system kernel or the application program.
The third type of technology inherits the traditional concept of data transmission between the address space of the application program and the address space of the operating system kernel, and then optimizes the data transmission itself. When data needs to be transferred between the buffer of the user address space and the page cache of the Linux operating system kernel, this technology can effectively improve the efficiency of data transfer.
There are many types of zero-copy technologies in Linux, and different technologies have their own advantages and disadvantages in different scenarios. Therefore, when choosing a zero-copy technology, it needs to be adjusted according to the specific situation.