The demo of sequence number generation in Cloud Spanner

Briefing

The demo uses Twitter snowflake id to generate sequence number in Cloud Spanner, which helps to smooth the migration from RDBMS with the sequence to Cloud Spanner.

Snowflake id is open source solution for generating unique ID numbers at high scale.

The demo uses the Java version of Twitter snowflake id.

Snowflake id description

The snowflake id is made up of these following components:

• Cont = 0 1 bit
• Epoch timestamp in millisecond precision - 41 bits
• Configured machine id - 10 bits
• Sequence number - 12 bits

The snowflake id is basically incremental, but cannot guarantee the id is consecutive.

How to run the demo

1. Create the table Orders in the Cloud Spanner

CREATE TABLE Orders (
    OrderId INT64,
    UserName STRING(MAX),
    Amount FLOAT64,
    OrderTime TIMESTAMP,
) PRIMARY KEY (OrderId)

2. Run the demo

 mvn compile exec:java -Dexec.mainClass=com.google.cloud.binguo.demo.App

3. Check the result in the Cloud Spanner console

select * from Orders;

The sequence number looks like below

579604939769401344
579604940381769728
579604940381769729
579604940381769730
579604940385964032
579604940385964033

Suggestion

Demo is not used for production. Before using sequence in Cloud Spanner, be noted that sequence should not be the prefix column of primary key, which causes hotspot and impact the performance.