What are you using CockroachDB for?
Exploring the feasibility of replacing a production system’s primary datastore with Cockroach (currently using Cassandra).
What convinced you to use CockroachDB?
I’ve been following the development of CockroachDB for several years, and have a strong interest in it as a scalable database system with strong consistency and referential integrity.
What languages are you using in your application stack?
Current language stack is Node.js, tho Go and Scala are also being explored as alternatives.
Is this for a personal side project or for a project at work? Would you consider using us at work?
Project at work, tho currently the exploration into Cockroach has been a personal endeavor. I’d love to try Cockroach at work, but it’s a hard sell before a v1 is released.
What has been the most pleasant surprise with CockroachDB?
Ease of deployment and maintenance, tho y’all really need to do fewer releases requiring a ‘stop the world’ deployment (just kidding, I know that until the v1 is out there’s really no expectation of such behavior)
What are some areas where CockroachDB was hard to use?
I know there’s some pride in the auto-documentation of SQL using grammar diagrams, but I find the flow-chart style diagrams more confusing than actual real-world examples. Especially when it comes down to having to follow links several times to see what is a valid input for a particular command. For instance, if I want to see what privileges I can grant to a user, I first go to the
grant_stmt documentation, then need to follow the
privileges link, which contains part of the potential privilege options (
ALL), otherwise I need to drill down into
privilege_list, which is itself a (from a user persective) useless section, only used to get to
privilege, which has the rest of the privileges I may want to grant someone.
GRANT isn’t meant to be a specific pain point, but rather just to illustrate the types documentation rabbit holes I’ve found myself digging through.
What features would make your experience better? Which ones are must-have?
Higher-order structured column types (equivalnts to Cassandra’s
map, and Postgres’
jsonb types. The latter (which can replace the behavior of the former) would be particularly helpful for some of the core use cases I have.