ShardingSphere Integration with CosId

Background

In the evolution of software systems, as business scale grows (TPS/storage capacity), we need to deploy applications in clusters to share computing and storage pressure. The stateless design of application services enables them to be scalable. When deploying with Kubernetes, we can complete service scaling with just one command (kubectl scale --replicas=5 deployment/order-service).

However, databases are not that easy when they become stateful. Database sharding becomes the obvious bottleneck for system performance.

Database Sharding

From a microservices perspective, vertical splitting is essentially microservice splitting. Defining service boundaries with bounded contexts splits large services/monolithic applications into multiple autonomous smaller-grained services. Because of the autonomy requirement, databases also need to be split by business. But after vertical splitting, individual microservices still face TPS/storage capacity challenges, so here we focus on horizontal splitting approaches.

Database sharding solutions are logically unified and physically partitioned. The core design lies in the middle-layer mapping scheme (Mapping in the figure above), i.e., the design of sharding algorithms. Almost all programming languages have built-in hash tables (java:HashMap/csharp:Dictionary/python:dict/go:map...). Sharding algorithms are highly similar to hash tables (hashCode), mapping key/shardingValue to corresponding slots (slot).

So where does shardingValue come from? CosId!!!

CosId: Distributed ID Generator

CosId aims to provide universal, flexible, high-performance distributed ID generators. CosId currently provides three algorithms:

SnowflakeId: Single-machine TPS performance: 409W/s, mainly solves clock backward issues, machine ID allocation issues, and provides more user-friendly and flexible usage experience.
SegmentId: Obtains a segment (Step) of IDs each time to reduce network IO request frequency of segment distributors and improve performance. Provides multiple storage backends: relational databases, Redis, Zookeeper for users to choose.
SegmentChainId(Recommended): SegmentChainId (lock-free) is an enhancement to SegmentId. Performance can reach near AtomicLong TPS performance: 12743W+/s.

The shardingValue problem is solved, but is that enough? ShardingSphere!!!

Excerpt from CosId official website: https://github.com/Ahoo-Wang/CosId

ShardingSphere

Apache ShardingSphere is an open-source distributed database ecosystem project consisting of JDBC, Proxy, and Sidecar (planned) products. It adopts a pluggable architecture, extending functionality through components. It provides various enhanced functions for databases and SQL on the upper layer, including data sharding, access routing, data security, etc.; on the lower layer, it natively supports multiple data storage engines such as MySQL, PostgreSQL, SQL Server, Oracle. The Apache ShardingSphere project philosophy is to provide database enhancement computing services, and build an ecosystem around it. By fully utilizing the computing and storage capabilities of existing databases, core capabilities are enhanced through plugin methods to help enterprises solve various difficulties encountered in digital transformation and accelerate digital application development.

Excerpt from Apache ShardingSphere official website: https://shardingsphere.apache.org/index_zh.html

Next, the main content of this article: How to integrate CosId based on ShardingSphere's pluggable architecture (SPI), and application configuration guide.

Installation

Taking Spring-Boot application as an example

ShardingSphere v5.1.0+

Because ShardingSphere v5.1.0 PR has merged cosid-shardingsphere, you only need to reference the ShardingSphere dependency.

xml

<dependency>
    <groupId>org.apache.shardingsphere</groupId>
    <artifactId>shardingsphere-jdbc-core-spring-boot-starter</artifactId>
    <version>5.1.1</version>
</dependency>

ShardingSphere v5.0.0

xml

<dependency>
    <groupId>org.apache.shardingsphere</groupId>
    <artifactId>shardingsphere-jdbc-core-spring-boot-starter</artifactId>
    <version>5.0.0</version>
</dependency>
<dependency>
    <groupId>me.ahoo.cosid</groupId>
    <artifactId>cosid-shardingsphere</artifactId>
    <version>1.8.15</version>
</dependency>

Distributed ID

KeyGenerateAlgorithm

UML Class Diagram

The figure shows all current ShardingSphere built-in KeyGenerateAlgorithm implementations. Here we only discuss CosIdKeyGenerateAlgorithm, please read https://shardingsphere.apache.org/document/current/cn/features/sharding/concept/key-generator/ for other implementations.

CosIdKeyGenerateAlgorithm

Configuration

type: COSID

Name	Data Type	Description	Default Value
id-name	`String`	Name of the `IdGenerator` (registered in `IdGeneratorProvider`)	`__share__`
as-string	`String`	Whether to generate string-type IDs	`false`

yaml

spring:
  shardingsphere:
    rules:
      sharding:
        key-generators:
          cosid:
            type: COSID
            props:
              id-name: __share__

Sharding Algorithms

ShardingAlgorithm

UML Class Diagram

CosIdModShardingAlgorithm

CosId modulo sharding algorithm

Algorithm Description

Single-value sharding key (PreciseShardingValue) algorithm complexity: O(1).
Range-value sharding key (RangeShardingValue) algorithm complexity: O(N), where N is the number of range values.

Performance Benchmark

Precise Value/Single Value (PreciseShardingValue)	Range Value/Multi Value (RangeShardingValue)

Configuration

type: COSID_MOD

Name	Data Type	Description	Default Value
logic-name-prefix	`String`	Logical table/datasource name prefix
mod	`int`	Divisor

yaml

spring:
  shardingsphere:
    rules:
      sharding:
        sharding-algorithms:
          alg-name:
            type: COSID_MOD
            props:
              mod: 4
              logic-name-prefix: t_table_

CosIdIntervalShardingAlgorithm

Interval-based time range sharding algorithm.

Algorithm Description

Precise value/single value sharding key (PreciseShardingValue) algorithm complexity: O(1).
Range value/multi value sharding key (RangeShardingValue) algorithm complexity: O(N), where N is the number of range value time units.

Performance Benchmark

Precise Value/Single Value (PreciseShardingValue)	Range Value/Multi Value (RangeShardingValue)

Configuration

type: COSID_INTERVAL

Name	Data Type	Description	Default Value
logic-name-prefix	`String`	Logical table/datasource name prefix
datetime-lower	`String`	Lower bound of time sharding, timestamp format: `yyyy-MM-dd HH:mm:ss`
datetime-upper	`String`	Upper bound of time sharding, timestamp format: `yyyy-MM-dd HH:mm:ss`
sharding-suffix-pattern	`String`	Sharding real table/datasource suffix pattern
datetime-interval-unit	`ChronoUnit`	Sharding key time interval unit
datetime-interval-amount	`int`	Sharding key time interval
ts-unit	`String`	Timestamp unit: `SECOND`/`MILLISECOND`	`MILLISECOND`
zone-id	`String`	Sharding key time zone	`ZoneId.systemDefault().getId()`

yaml

spring:
  shardingsphere:
    rules:
      sharding:
        sharding-algorithms:
          alg-name:
            type: COSID_INTERVAL
            props:
              logic-name-prefix: logic-name-prefix
              datetime-lower: 2021-12-08 22:00:00
              datetime-upper: 2022-12-01 00:00:00
              sharding-suffix-pattern: yyyyMM
              datetime-interval-unit: MONTHS
              datetime-interval-amount: 1

CosIdSnowflakeIntervalShardingAlgorithm

Algorithm Description

We know the bit partitioning method of SnowflakeId, SnowflakeId can parse out the timestamp, that is, SnowflakeId can be used as time, so SnowflakeId can be used as the sharding value for the INTERVAL sharding algorithm. (When there is no CreateTime available for sharding [this is an extremely rare case], or when there are extremely demanding performance requirements, using distributed ID primary key as the query range may be a better choice for persistence layer performance.)

Configuration

type: COSID_INTERVAL_SNOWFLAKE

Name	Data Type	Description	Default Value
logic-name-prefix	`String`	Logical table/datasource name prefix
datetime-lower	`String`	Lower bound of time sharding, timestamp format: `yyyy-MM-dd HH:mm:ss`
datetime-upper	`String`	Upper bound of time sharding, timestamp format: `yyyy-MM-dd HH:mm:ss`
sharding-suffix-pattern	`String`	Sharding real table/datasource suffix pattern
datetime-interval-unit	`ChronoUnit`	Sharding key time interval unit
datetime-interval-amount	`int`	Sharding key time interval
id-name	`String`	Name of the `IdGenerator` (registered in `IdGeneratorProvider`)	`__share__`

yaml

spring:
  shardingsphere:
    rules:
      sharding:
        sharding-algorithms:
          alg-name:
            type: COSID_INTERVAL_SNOWFLAKE
            props:
              logic-name-prefix: logic-name-prefix
              datetime-lower: 2021-12-08 22:00:00
              datetime-upper: 2022-12-01 00:00:00
              sharding-suffix-pattern: yyyyMM
              datetime-interval-unit: MONTHS
              datetime-interval-amount: 1
              id-name: cosid-name

Summary

This article mainly discusses the background of database sharding and how to integrate CosId based on ShardingSphere's pluggable architecture for practical applications. ShardingSphere adopts a pluggable architecture, making it very convenient for developers to customize functions that meet their application scenarios. If you are also interested in contributing to the ShardingSphere community, please refer to https://shardingsphere.apache.org/community/cn/contribute/contributor/.

Reading Source Code Tips: Class Diagrams

Many partners get lost in the details when reading source code, traversing line by line based on methods, getting lost in the details (if you can persist, I admire your tenacity!). This reading method is very bad and inefficient. Reading source code is very similar to reading books: first establish an overview map (index), then drill down layer by layer. (Top-down approach is more conducive to not getting lost in specific details during reading) We recommend using the IntelliJ IDEA plugin Diagrams to generate source code level overview diagrams: UML class diagrams.

IntelliJ IDEA: https://www.jetbrains.com/help/idea/class-diagram.html

References

ShardingSphere Official Documentation: https://shardingsphere.apache.org/document/current/cn/overview/
IntelliJ IDEA: https://www.jetbrains.com/help/idea/class-diagram.html
CosId-ShardingSphere: https://cosid.ahoo.me/guide/cosid-shardingsphere.html

ShardingSphere Integration with CosId ​

Background ​

Database Sharding ​

CosId: Distributed ID Generator ​

ShardingSphere ​

Installation ​

Distributed ID ​

UML Class Diagram ​

CosIdKeyGenerateAlgorithm ​

Configuration ​

Sharding Algorithms ​

UML Class Diagram ​

CosIdModShardingAlgorithm ​

Algorithm Description ​

Performance Benchmark ​

Configuration ​

CosIdIntervalShardingAlgorithm ​

Algorithm Description ​

Performance Benchmark ​

Configuration ​

CosIdSnowflakeIntervalShardingAlgorithm ​

Algorithm Description ​

Configuration ​

Summary ​

Reading Source Code Tips: Class Diagrams ​

References ​

ShardingSphere Integration with CosId

Background

Database Sharding

CosId: Distributed ID Generator

ShardingSphere

Installation

Distributed ID

UML Class Diagram

CosIdKeyGenerateAlgorithm

Configuration

Sharding Algorithms

UML Class Diagram

CosIdModShardingAlgorithm

Algorithm Description

Performance Benchmark

Configuration

CosIdIntervalShardingAlgorithm

Algorithm Description

Performance Benchmark

Configuration

CosIdSnowflakeIntervalShardingAlgorithm

Algorithm Description

Configuration

Summary

Reading Source Code Tips: Class Diagrams

References