Shulou(Shulou.com)06/01 Report--

This article mainly explains the "PHP function uniqid () can generate a unique ID", the article explains the content is simple and clear, easy to learn and understand, the following please follow the editor's ideas slowly in depth, together to study and learn "PHP function uniqid () can generate a unique ID" bar!

Although PHP provides a function uniqid () that generates a unique ID, can this function really generate a unique ID? Let's take a look at the implementation of uniqid ():

PHP_FUNCTION (uniqid)

{

...

Gettimeofday ((struct timeval *) & tv, (struct timezone *) NULL)

Sec = (int) tv.tv_sec

Usec = (int) (tv.tv_usec% 0x100000)

Spprintf (& uniqid, 0, "sxx", prefix, sec, usec)

RETURN_STRING (uniqid, 0)

}

As you can see from the code, uniqid () is implemented through subtle timestamps, and in the case of distributed high concurrency, the repetition rate of ID is very high, so we cannot use uniqid () to generate a unique ID.

Snowflake algorithm

Since uniqueness cannot be guaranteed by timestamp alone, can the following eigenvalues be added to ensure uniqueness? For this reason, Twitter invented the snowflake algorithm. The core principle of the snowflake algorithm is to divide a 64-bit integer into three parts, as shown in the following figure:

As shown in the figure above, the first bit of the high end is not used, and the next 41 bits are used to store the millisecond timestamp, followed by 10 bits of the timestamp as the machine ID, and the last 12 bits as the serial number.

For different machines, a unique machine ID can be assigned to each machine, which ensures that the ID generated by each machine lock will not be duplicated.

For the same machine, if multiple clients request concurrently at the same time, ID uniqueness can be ensured by increasing the sequence number.

By default, a 41-bit timestamp can support the algorithm until 2082 (by subtracting a starting timestamp), a 10-bit working machine ID can support 1023 machines, and a 12-bit sequence number supports 4095 self-increasing sequence ID in 1 millisecond.

In other words, a machine can withstand 4095000 concurrency per second, which is suitable for any scenario. The code implementation of the snowflake algorithm is roughly as follows:

Static uint64_t last_ts = 0

Static uint64_t sequence = 0

Static uint64_t datacenter_id = 0

/ / get millisecond timestamp

Uint64_t current_timestamp ()

{

Struct timeval tv

Uint64_t retval

If (gettimeofday (& tv, NULL) =-1) {

Return 0ULL

}

Retval = (u64buttt) tv.tv_sec * 1000ULL +

(u64buttt) tv.tv_usec / 1000ULL

Return retval

}

/ / if the concurrency now is exceeded in the same millisecond, wait for the next millisecond

Uint64_t skip_next_millis ()

{

Uint64_t ts

While (1) {

Ts = current_timestamp ()

If (ts > last_ts) {

Break

}

}

Return ts

}

/ / get the next ID

Uint64_t get_next_id ()

{

Uint64_t retval, ts

Ts = current_timestamp ()

If (ts = = last_ts) {/ / multiple concurrency in the same millisecond

Sequence = (sequence + 1) & 0xFFF; / / add serial number counter

If (sequence = = 0) {/ / counter is used up

Ts = skip_next_millis (); / / wait for the next millisecond

}

} else {

Sequence = 0; / / clear the serial number counter

}

Last_ts = ts

Retval = (ts

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.