workspace-images VS ybdemo

If your project uses PostgreSQL, you can test it on YugabyteDB (Open Source PostgreSQL-compatible Distributed SQL database) with the dedicated YugabyteDB image. Simply base your .gitpod.dockerfile on:

As I did with Oracle, I can run a single insert and look at the statistics from explain analyze and my ybwr:

I've also updated my YugabyteDB Lab with Grafana for this scenario, running:

I check, with my ybwr.sql script, that reading one row from the job_fanout view reads from one table only:

Here is a quick test using my ybwr on this demo table which has 3 tablets across 3 nodes.

If you want to play with a multi-node, as you need multiple network interfaces, it is better not to use this image but start YugabyteDB nodes in docker, as I do in my ybdemo:

@franckpachot

I've used this docker-compose

I'm starting a RF=3 cluster on my laptop using my ybdemo/docker/yb-lab/ and I set two aliases, ysqlsh, for SQL commands, and yb-admin, for snapshot commands:

I also load my ybwr.sql to show the number of rocksdb seek() and next() in the tservers:

Now that you know how to scale out, just with the --join option of yugabyted, you can build your docker-compose file. For example, to replace PostgreSQL in a Docker Compose used for tests, I use the same variables as the PostgreSQL image - example here. And yugabyted is convenient for quick start, but you can have more control by starting all components like I do when I demo high availability and elasticity: https://github.com/FranckPachot/ybdemo/tree/main/docker/yb-lab