Our use cases are for data analytics, both real-time and batch, and also for logging Clickstream data.

We use it in our organization. We have it in our production environment.

The query engine is super fast. We deploy ClickHouse on our Kubernetes cluster, not as a cloud subscription, so it's easy to scale with the deployment.

Some features, like connecting to third-party applications or the cloud, could be better.

I have been using it for one year.

One issue is that you need persistent volumes. Otherwise, if one system goes down, you lose data in that cluster.

Another issue is performance. You have to make sure you have the right configurations; otherwise, it will lead to queuing where all your jobs get queued.

It is a scalable product.

You only get technical support when you take the cloud subscription. If you have it in-house, you won't get any support. If you have a cloud subscription, then the support is pretty good. You can raise a ticket from the UI, and they will respond within 24 hours.

So, the support team is pretty good but there is a little room for improvement.

The initial setup is pretty difficult since we deployed it in-house. We didn't use the cloud subscription, so we have to handle the deployment very carefully.

The challenge was deploying it and having the replication concept working. Another challenging feature is persistent volumes. You have to make sure the data is available on all clusters; otherwise, if one cluster goes down, you'll lose all your data. It's better to have it replicated.

We first used the cloud subscription, but we saw a possibility to reduce costs, so we tried deploying the open-source ClickHouse on-premises. That saved us money, but we didn't get all the features that come with the subscription.

We did it in-house.

Pricing for the cloud version is alright, not very costly or cheap.

But if you have an in-house deployment on Kubernetes or something, it's going to be very cheap since you'll be managing everything.

I would tell other users to do a POC because it depends upon the business use case and the data. They can explore first. There's another open-source option called Apache Druid, which is a little better than ClickHouse. If that doesn't fit the use case, then they could go for ClickHouse.

Overall, I would rate the solution a seven out of ten.

If you have a real-time basis, you should take a look at ClickHouse because it works on a vector database, and the querying is super fast compared to traditional databases. So, if your use case is real-time or logging or real-time dashboarding, then ClickHouse is a tool to consider. Otherwise, if it's batch processing and you can expect some latency, then you should go for other databases.