I am building an automation test suite which is running on multiple machines, all reporting their status to a postgresql database. We will run a number of automated tests for which we will store the following information:

• test ID (a GUID)
• test name
• test description
• status (running, done, waiting to be run)
• progress (%)
• start time of test
• end time of test
• test result
• latest screenshot of the running test (updated every 30 seconds)

The number of tests isn't huge (say a few thousands) and each machine (say, 50 of them) have a service which checks the database and figures out if it's time to start a new automated test on that machine.

How should I organize my SQL table to store all the information? Is a single table with a column per attribute the way to go?

If in the future I need to add attributes but want to keep compatibility with old database format (ie I may not want to delete and create a new table with more columns), how should I proceed? Should the new attributes just be in a different table?

I'm also thinking of replicating the database. In case of failure, I don't mind if the latest screenshots aren't backed up on the slave database. Should I just store the screenshots in its own table to simplify the replication?

Thanks!