I’m a huge fan of Blazer. At a previous company, we always ran it against a read-only DB replica (which we already had) to avoid having to deal with user permissions, but I do like the idea of Hypershield filtering out certain columns. That looks like it requires making that secondary user, right?
The other thing about Blazer that I always loved was that it helped us check a given query’s performance against a real production dataset. When you run a query against a development dataset, it’s always going to be fast because there’s next to nothing there, and you also can’t trust the query planner to run a given query the same way in two different DBs since it adapts to the quantity and diversity of the data. Running the actual query plan in production is super nice.
I’m a huge fan of Blazer. At a previous company, we always ran it against a read-only DB replica (which we already had) to avoid having to deal with user permissions, but I do like the idea of Hypershield filtering out certain columns. That looks like it requires making that secondary user, right?
The other thing about Blazer that I always loved was that it helped us check a given query’s performance against a real production dataset. When you run a query against a development dataset, it’s always going to be fast because there’s next to nothing there, and you also can’t trust the query planner to run a given query the same way in two different DBs since it adapts to the quantity and diversity of the data. Running the actual query plan in production is super nice.
Yes, because you don't want to restrict your default user to anything otherwise you will probably have a bad time haha