Why is nginx so fast?

The main reason is that Nginx uses the latest epoll (Linux 2.6 kernel) and kqueue (freebsd) network I/O model.

epoll is a method of multiplexing IO (I/O Multiplexing), but it is only used for Linux2.6 and above kernels. Before we start discussing this issue, let us first explain why. Multiplexing IO is required.

Let’s explain with an example in life.

Suppose you are studying in college and have to wait for a friend to visit, and this friend only knows that you are at No. A Building, but I don’t know where you live specifically, so you made an appointment to meet at the door of Building A.

If you use the blocking IO model to deal with this problem, then you can only wait at Building A all the time. Waiting for the arrival of your friends at the door of Building No. 1. During this time, you cannot do anything else. It is not difficult to know that the efficiency of this method is low.

Now times have changed, and multiplexing has begun to be used. The IO model is used to deal with this problem. You tell your friend to come to Building A and ask the building manager to tell you how to go. The building manager here plays the role of multiplexed IO.

Further explain the difference between the select and epoll models.

The select version of the aunt does the following: For example, when classmate A’s friend comes, the select version of the aunt is stupid, and she takes her friends from room to room to check who is who. It’s classmate A. The friend you were waiting for has arrived, so in the actual code, the select version of the aunt does the following:

##int n = select(&readset,NULL,NULL,100);

for (int i = 0 ; n > 0; i) {

if (FD_ISSET(fdarray[i], &readset)){
do_something( fdarray[i]);
--n;
}
}

##epoll version of aunt is more advanced, she wrote down classmate A information, such as his room number, then when classmate A's friend arrives, you only need to tell the friend which room classmate A is in, and you don't have to go around the building full of people to find someone. So the epoll version of the aunt did Things can be represented by the following code:

##Code highlighting produced by Actipro CodeHighlighter (freeware)

http://www.CodeHighlighter.com/
n

=epoll_wait(epfd,events,20,500);
for
(i=0;i<n; i){do_something(events[n]) ;}

In epoll, the key data structure epoll_event is defined as follows:

typedef union epoll_data {
void * ptr;
int fd;__uint32_t u32;
__uint64_t u64;
} epoll_data_t;


struct epoll_event {__uint32_t events;
/* Epoll events */epoll_data_t data ;
/* User data variable */};

You can see, epoll_data is a union structure, which is the structure used by the epoll version of aunt to save classmate information. It can save many types of information: fd, pointers, etc. With this structure, epoll aunt can locate it without any effort. To classmate A.

Don’t underestimate these efficiency improvements. In a large-scale concurrent server, polling IO is one of the most time-consuming operations. Going back to the example, if every A friend's building manager wants to query the classmates in the whole building, so the processing efficiency will inevitably be low. Soon there will be a lot of people on the floor.

Compare the earliest blocking IO processing model , it can be seen that after using multiplexed IO, the program can freely perform its own work except IO operations. Only when the IO status changes, multiplexed IO will notify it, and then take corresponding operations. , instead of having to block and wait for the IO status to change.

It can also be seen from the above analysis that the improvement of epoll over select is actually a specific application of the idea of ​​​​exchanging space for time.

For more Nginx related technical articles, please visit the Nginx Tutorial column to learn!

The above is the detailed content of Why is nginx so fast?. For more information, please follow other related articles on the PHP Chinese website!