What is the typical hardware cost for deploying a production environment TiDB for core business with a data volume of 40TB?

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

Original topic: 部署一套生产环境TiDB硬件成本一般需要多少,核心业务,数据量40TB

| username: alfred

【TiDB Usage Environment】Production\Test Environment\POC
【TiDB Version】
【Encountered Issue】
【Reproduction Path】What operations were performed that led to the issue
【Issue Phenomenon and Impact】

| username: cs58_dba | Original post link

There’s no significant difference between a server and a regular database, except that NVE hard drives are a bit more expensive.

| username: BinLi1988 | Original post link

40TB requires 20 pieces of 1.96T NVMe or regular SSDs.

  1. Deploy 4 TiKV instances, requiring 5 TiKV nodes.
  2. This setup should be accompanied by 3 TiDB servers and 3 PD nodes, or a mixed deployment.
    Back in 2019, we spent 700,000 RMB on 6 servers.
    Among them, 3 servers had 6 NVMe drives each, with 2 drives left unused, and the remaining 3 servers only had system disks for deploying TiDB servers, all with 80 cores and 250GB of memory.

Now, this setup is estimated to cost less than 500,000 RMB, and it would be even cheaper with regular SSDs.

| username: TiDBer_ga8bJIx0 | Original post link

You haven’t considered disaster recovery.

| username: alfred | Original post link

Disaster recovery and offline backup need to be considered.

| username: jansu-dev | Original post link

This question is quite broad:

  1. Adaptation and migration costs;
  2. Hardware costs to meet performance requirements;
  3. Whether disaster recovery is needed, and what kind of disaster recovery plan;
  4. And so forth…

You can’t just estimate based on the maximum disk capacity and the number of disks needed. If you consider it more comprehensively, you also need to consider the cost of maintenance personnel, etc.
It is recommended to contact PingCAP’s official sales and technical team to develop a feasible, business-requirement-meeting, and low-risk plan.

| username: 张雨齐0720 | Original post link

You need to provide specific disaster recovery requirements. Same city? Remote disaster recovery? There are too many machine resource specifications to plan for each one. Provide some prerequisites so that an assessment can be made.