I would like CockroachDB to have more compatibility with PostgreSQL, especially with the connection string and technical integrations.

I find the serverless offer a bit confusing. The price is a function of the maximum database size.

A single SELECT can use a number of request units, for example. If the throughput is high enough, you could run out of request units and be throttled if you do not pay attention to how your database is used.

CockroachDB's automatic failover capability is an area of concern where improvements are required.

The product is not an enterprise-level solution. The product must improve its disaster recovery features. It must also improve on support, logging, and backup restore.

I think the engine itself could be improved. If you miss a partition key, it should be able to be modeled. This is where AI could be really helpful. For example, if a user is from Australia, the system should know to look for their data there first instead of searching in the US or another continent.

The platform could be more extensible. At the moment, I have to use Postgres for extension issues. There needs to be better integration features. Additionally, the product’s Change Data Capture Integrations with tools such as Kafka needs improvement.

CockroachDB needs to improve store processes. 

The closer they can make CockroachDB to being completely compatible with Postgres, the better. It's almost compatible, but not completely. If it was, it would be nice to just be able to use Postgres libraries without any fiddling.

Cockroach does not support all types of protocols. I need to improve it myself to support a CouchDB on my network.

We are looking for more features to support distributed high availability and geo-partitioning.

As additional features, I wish there was Japanese documentation, instead of just English. I know how powerful and simple CockroachDB can be for many Japanese people with the correct documentation.

The initial setup and pricing could be improved.