Hey, This is the last chapter about storage management of the TimescaleDB course. After reading this one, you know everything to ensure your analytics will never have any issues because of data size. So, retention policies from the last chapter are a nice way to drop old data you no longer need and save costs. But can you absolutely be sure that you never need that data again? I don't know how you feel, but deleting something always makes me a little bit nervous.

That's exactly why you have Tiered Storage as an alternative solution in TimescaleDB: Instead of deleting data, it is moved to cloud storage (outside your database) to reduce storage costs. The data still exists and can be queried if needed, but it won't be as fast as your TimescaleDB database. So you can effectively optimize your storage by combining high-performance storage (your TimescaleDB database) with low-cost storage (object storage). Again, with a simple policy to automatically move chunks with all data at least e.g. 3 months old.

Because tiered storage is (a lot) slower than your TimescaleDB database, tiered chunks are not included in queries by default. And why should they? After all, those are the ones you declared as no longer needed, so no query uses them. But when you decide you want to run a one-off query that includes the old data, you can do so - with a performance impact. Or you untier those chunks to move them back into your TimescaleDB database anytime you want.

The cost savings of tiered storage multiply really fast: You're most likely running your database in a high-availability setup with at least one copy and four replicas, because you have many users and two active forks by developers testing new features. You would be paying for eight times the storage. But with tiered storage, all tiered chunks are billed only once (!) because they are stored only once in object storage, not duplicated across every database.

That's a rough overview of the course's chapter but it describes the main idea. Learn about tiered storage and the cost savings you can achieve for your TimescaleDB database. If that sounds interesting to you, please take a look at it and let me know what you think!

I hope you learn a lot
Tobias