> The latter part of that workflow is where we are focused now, i.e. now that we can plan and recommend B-tree indexes for individual queries, how do we utilize this to recommend a set of indexes for a whole database
what I've ended up doing is trying to approximate the number of rows, create dummy data that has a similar spread of data, and create indexes and query until I have a good understanding of exactly how each index behaves with the amount of data. It ends up being a lot of work, but sometimes getting a 50% performance increase is worth that amount of work overall.
> On open-sourcing this code: Not at this moment
completely understandable, the selfish side of me hopes that it will be open sourced eventually, but the practical side of me sees the benefits to your company for keeping it private for now.
what I've ended up doing is trying to approximate the number of rows, create dummy data that has a similar spread of data, and create indexes and query until I have a good understanding of exactly how each index behaves with the amount of data. It ends up being a lot of work, but sometimes getting a 50% performance increase is worth that amount of work overall.
> On open-sourcing this code: Not at this moment
completely understandable, the selfish side of me hopes that it will be open sourced eventually, but the practical side of me sees the benefits to your company for keeping it private for now.