Can all TiDB components be scaled from a single node to a cluster environment?

Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: TiDB所有组件可以从单节点扩容到集群环境吗?

| username: ohammer

[TiDB Usage Environment] Production Environment / Testing / PoC
[TiDB Version]
[Reproduction Path] What operations were performed when the issue occurred
[Encountered Issue: Issue Phenomenon and Impact]
[Resource Configuration]
[Attachments: Screenshots / Logs / Monitoring]

Currently, we want to use TiDB as a data middle platform, and it is still in the testing phase. Due to limited public cloud resources, we want to deploy all components on a single server first to verify the feasibility of data synchronization. Once verification is complete, we will apply for resources to expand into a multi-node cluster environment. Is this feasible?

| username: xfworld | Original post link

In that case, I would suggest you apply for at least three nodes: one node for PD, one node for TiDB, and one node for TiKV. You can temporarily skip installing other components. Additionally, you will need a control machine to execute tiup, making a total of four nodes.

This setup will barely run… With only a single TiKV node, it won’t meet the requirement for three replicas…

If you need to scale up later, you can add more nodes and continue scaling up using tiup.

| username: 啦啦啦啦啦 | Original post link

The method is feasible, but wouldn’t it be troublesome to scale up and down later? It feels like it would be better to redeploy a new set after verifying the solution.

| username: Billmay表妹 | Original post link

Here are some practical cases of data middle platforms using TiDB that you can refer to:

Data Middle Platform

Column - TiDB in Chain Fast Food Enterprises丨Exploration of Massive Transactions and Real-time Analysis Applications

Column - Dry Goods | Distributed Database TiDB Practice in Ctrip

Yiguo’s Use of TiDB and Thoughts on Data Middle Platform

Beike Financial TiDB Online Cross-data Center Migration Practice

Digital Acceleration, How to Keep Data Fresh and Realize Data Value? — TiDB Industry Application Scenarios and Case Interpretation

Selected Practices: TiDB Implementation in New Oriental’s Business Front and Middle Platforms

Real-time Data/Business Analysis

Column - TiDB HTAP Meets New Energy Vehicle Enterprises: Application Practice of Real-time Data Analysis in Direct Sales Mode

Column - TiDB in Ctrip | Real-time Tag Processing Platform Optimization Practice

Column - TiDB Implementation in Getui | Solving Hotspot Problems and Improving Performance by Over a Thousand Times

Column - Technology Upgrade & Industry Upgrade TiDB Helps Yiche Create a Super Car Carnival

Column - Dry Goods Sharing丨Ctrip International Business Dynamic Real-time Tag Processing Platform Practice

Column - TiDB Application Practice in iQIYI Real-time Analysis Scenarios

| username: tidb菜鸟一只 | Original post link

You can deploy multiple nodes on one machine. For example, you can deploy 3 TiKV, 1 PD, and 1 TiDB on a single machine. If you need to scale out later, you can add more nodes on other machines and then scale down the local machine nodes, using it as the TiUP control machine. However, this setup can only be used for functional testing, not performance testing.

| username: ffeenn | Original post link

It is recommended to use SSDs or hard drives with higher read and write capabilities for cloud storage. The higher, the better.

| username: ohammer | Original post link

Since applying for public cloud servers requires costs and the company has a state-owned background, it is very difficult to apply for expenses. We want to directly use the existing resources.

| username: ohammer | Original post link

Makes sense, I’ll try it out soon.

| username: system | Original post link

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.