When a system is small, MySQL works very well.

But as a business grows, data volume increases, traffic rises, and scaling requirements become more demanding, MySQL eventually starts to hit its limits.

While exploring distributed database solutions, I had the opportunity to learn about TiDB. It is an interesting database system: MySQL-compatible, yet designed for horizontal scalability from the beginning.

In this blog, I share what I learned about TiDB to help the team better understand this solution and to evaluate whether it fits our future projects.

What Is TiDB?

TiDB is an open-source distributed SQL database, developed by PingCAP since 2015.

Its most important characteristics are:

• Familiar SQL interface

• Horizontal scalability

• No manual sharding

Easy like MySQL — Powerful like NoSQL

TiDB Architecture Overview

TiDB is built as a distributed system with four core components.

1. TiDB Server — SQL Layer

TiDB Server is responsible for SQL processing:

• Receives queries from applications

• Parses and optimizes SQL

• Coordinates data reads and writes

TiDB Server does not store data.

It can be seen as the "MySQL layer" in a distributed system.

2. TiKV — Row-based Storage

TiKV is the primary storage layer:

• Uses a key-value data model

• Optimized for OLTP

• Data is distributed across multiple nodes

Each table is split into regions and automatically balanced within the cluster.

3. TiFlash — Column-based Storage

TiFlash is a column-based storage engine:

• Designed for OLAP

• Used for analytics and reporting

• Optimized for large data scans

TiFlash enables TiDB to handle transactions and analytics in the same system.

4. PD (Placement Driver) — The "Brain" of the Cluster

PD is responsible for:

• Managing metadata

• Deciding where data should be placed

• Load balancing across nodes

• Managing replicas

PD ensures that the entire cluster operates stably and efficiently.

How TiDB Works (Execution Flow)

A typical SQL query in TiDB follows this flow:

Or step by step:

1. The application sends an SQL query.

2. TiDB Server parses and optimizes the query.

3. PD determines where the data is located.

4. TiKV or TiFlash processes the read/write.

5. The result is returned to the application.

Who Is Using TiDB?

Pinterest

In 2022, Pinterest evaluated multiple large-scale database technologies and found TiDB to deliver the most stable performance.

Shopee

Uses MySQL and TiDB in parallel. The number of TiDB clusters continues to grow for workloads that require rapid scaling.

Ninja Van

Operates more than 130TB of data and serves millions of queries per minute with only two engineers managing the infrastructure.

Zhihu

With data growth of over 100 billion records per month, standalone MySQL could no longer keep up.

Key Advantages of TiDB

MySQL-compatible

Most existing SQL syntax and drivers work without modification.

True Scale-out

Add nodes without downtime or manual sharding.

HTAP

Supports both OLTP and OLAP in a single system via TiFlash.

Flexible Deployment

Runs on:

• AWS

• GCP

• Azure

• On-premises

• Kubernetes

Open Source

No vendor lock-in and flexible cost optimization.

Trade-offs to Consider

More Complex to Operate than MySQL

Requires managing multiple components:

• PD

• TiKV

• TiDB Server

• TiFlash

Higher Latency than Local MySQL

Due to distributed reads and writes.

Not 100% MySQL Compatible

Some features are not supported such as:

• Triggers

• Stored Procedures

• Events

MySQL Compatibility

TiDB is MySQL-compatible, but not 100% compatible.

Some syntax, functions, and behavior differ (v8.5).

Documentation:

• https://docs.pingcap.com/tidb/stable/basic-features/

• https://docs.pingcap.com/tidb/stable/mysql-compatibility/

Cost Comparison: MySQL RDS vs TiDB Cloud

Try TiDB in a Local Environment

Step 1 — Install TiUP Playground

Open Terminal and run:

curl --proto'=https' --tlsv1.2 -sSf https://tiup-mirrors.pingcap.com/install.sh | sh`

After installation:

source ~/.bash_profile
# or
source ~/.zshrc

Check the version:

tiup --version

Step 2 — Start a Local TiDB Cluster

Run:

tiup playground --db 1 --pd 1 --kv 1 --tiflash 1 --monitor

After running the tiup playground command, TiUP provides several important URLs/endpoints:

TiDB (MySQL Endpoint)

`bash mysql --host 127.0.0.1 --port 4000 -u root `

This endpoint is compatible with MySQL and can be used by:

• Laravel

• DBeaver

• TablePlus

• MySQL Workbench

• Other MySQL clients
  

Example .env configuration in Laravel project:

DB_CONNECTION=mysql
DB_HOST=127.0.0.1
DB_PORT=4000
DB_DATABASE=demo
DB_USERNAME=root
DB_PASSWORD=

TiDB Dashboard

http://127.0.0.1:2379/dashboard

Example:

Grafana (Monitoring)

http://127.0.0.1:3000

Monitoring dashboards (QPS, latency, CPU, memory, TiKV/TiFlash metrics)
Default login: admin / admin
Example:

Scale Up the Cluster

You can easily start a larger cluster.

`bash tiup playground --db 2 --pd 2 --kv 3 --tiflash 3 --monitor `

Note: This setup is for the local environment. You can use TiDB Cloud for the production environment.

When Should You Consider TiDB?

TiDB is a good choice when your system has:

• Large and fast-growing datasets

• Unpredictable or spiky traffic

• Real scale-out requirements

• Avoiding manual sharding

• Need for real-time OLTP + OLAP

Conclusion

TiDB does not replace MySQL in every scenario.

However, when MySQL starts to reach its limits, TiDB becomes a strong alternative.

With horizontal scalability, a clear architecture, and a familiar SQL experience, TiDB provides a practical and long-term solution for growing systems.