TiDB Standalone Edition

Translated
1

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

Original topic: TiDB单体版

| username: xuexiaogang

I have an idea, but I’m not sure how feasible it is. OB has released version 4, which is a standalone version. I’m wondering if TiDB should also have a standalone version? Oracle has a standalone version, and you can scale it to a cluster (of course, the underlying layer needs to be changed to ASM, but that’s another issue).

Here’s why I’m suggesting this. I’ve always complained about poor SQL development, and as I interact with more users, even those from banks, I find that their SQL development is also poor. Perhaps only companies like Alibaba, Tencent, Huawei, and ByteDance have good SQL development. In other words, poor SQL development is an abnormal normality.

Many issues on the forums are not product issues but are caused by poor SQL. Some decision-makers in various companies are aware of the poor development skills but accept this reality, or they can’t change it. Given this, a large cluster, such as merging a dozen databases into TiDB, would also face issues if hit by poor SQL all at once.

Therefore, they are hesitant to do so and prefer to separate them (the downsides of separation are numerous, but let’s not discuss that today). If it can be divided into N standalone instances, the cost can be controlled. If it is divided into N clusters, even if the database is free, the servers and storage still cost money.

Moreover, many systems can indeed be handled by a standalone instance, so will there be a standalone version?

2
| username: ShawnYan | Original post link

It feels a bit difficult. OB can achieve all in observer, but TiDB starts as a distributed cluster.

3
| username: xuexiaogang | Original post link

However, this is indeed competitive. Think about it, many companies’ scenarios can be handled by a single machine. With more usage, the ecosystem and market will emerge.

4
| username: forever | Original post link

Distributed systems are designed to address the limitations of a single machine. If a single machine is not a bottleneck, then MySQL should suffice, and there would be no need for TiDB.

5
| username: ShawnYan | Original post link

So how do you view the single-machine performance of OB 4.0 surpassing MySQL? Comparison test between TiDB and MySQL on the same single-machine spec?