Distributed ID

Distributed ID What conditions need to be met ？

How to generate unique keys ：

1.UUID

2. Database auto increment primary key

3. Database cluster dual master mode

4. Segment mode based on Database

5. be based on Redis Pattern

6. Snowflake algorithm （Snowflake） Pattern

---

Distributed ID What conditions need to be met ？

• Globally unique ： Must ensure ID It's global and unique , The basic requirements
• High performance ： High availability and low delay ,ID Generate response blocks , Otherwise, it will become a business bottleneck
• High availability ：100% 's availability is deceptive , But also to be infinitely close to 100% The usability of
• Good access ： We should adhere to the design principle of "use as you go" , The system design and implementation should be as simple as possible
• The trend is increasing ： Best trend increasing , This requirement depends on the specific business scenario , Generally not strict

How to generate unique keys ：

1.UUID

2. Database auto increment primary key

Implement a simple , But there is a single point problem , Unable to resist high concurrency scenarios

3. Database cluster dual master mode

Set different starting values and step sizes for each database

Solve a single problem , But it is not conducive to subsequent expansion , It still can't solve the new concurrency problem

4. Segment mode based on Database

Segment mode is now distributed ID One of the main ways to implement generator , The number segment mode can be understood as obtaining auto increment from database batch ID, One segment range at a time from the database , for example (1,1000] representative 1000 individual ID, Specific business services , Generate 1~1000 Self increasing of ID And load it into memory .

5. be based on Redis Pattern

Redis It can also be achieved , The principle is to use redis Of incr Command implementation ID Atomic self increasing of .

6. Snowflake algorithm （Snowflake） Pattern

Snowflake The generated is Long Type of ID, One Long Type account 8 Bytes , Each byte takes up 8 The bit , That is to say a Long Type account 64 A bit .

Snowflake ID Make up the structure ： Positive digit （ Occupy 1 The bit ）+ Time stamp （ Occupy 41 The bit ）+ machine ID（ Occupy 5 The bit ）+ Data Center （ Occupy 5 The bit ）+ Self value added （ Occupy 12 The bit ）, in total 64 One of the bits Long type .

• first bit position （1bit）：Java in long The highest bit of is the sign bit representing the positive and negative , A positive number is 0, Negative number is 1, General generation ID All positive , So the default is 0.
• Time stamp part （41bit）： Time in milliseconds , It is not recommended to save the current time stamp , It's about using （ Current timestamp - Fixed start timestamp ） The difference between the , Can make ID Start with a smaller value ;41 Bit timestamps can be used 69 year ,(1L << 41) / (1000L 60 60 24 365) = 69 year
• Work the machine id（10bit）： It's also called workId, This can be configured flexibly , Machine room or combination of machine numbers .
• Serial number part （12bit）, Self increment supports the generation of the same node in the same millisecond 4096 individual ID

According to the logic of this algorithm , Just use this algorithm Java Language is realized , Encapsulation as a tool method , Then each business application can directly use the tool method to obtain the distributed ID, Just make sure that each business application has its own working machine id that will do , It doesn't need to build a single access distribution ID Application .