Why Your Database Isn't Scaling (And How to Fix It)

Your database will eventually stop scaling by getting bigger. The biggest instance your cloud provider sells is a hard ceiling — and once you hit it, the only way forward is out, not up. In this video we walk through database scaling end to end: vertical vs horizontal scaling, why multi-master databases have it easy and master-replica setups don't, and how sharding splits a single table across nodes. Then the part most explanations skip — how a badly chosen shard key creates hot shards, forces resharding, and leaves you maintaining a shard tree. Part of the System Design series. Previous video: choosing the right database for the right scenario →    • SQL or NoSQL? How to Pick Right Database E...   CHAPTERS 00:00 What this video covers 00:46 What "scaling" actually means 01:04 Vertical vs horizontal scaling 01:56 The hard ceiling of vertical scaling 02:40 Multi-master vs master-replica databases 03:51 Why SQL pushes you toward sharding 04:00 Sharding a users table, step by step 05:13 One table, three shards 06:07 Horizontal vs vertical sharding 07:18 Why horizontal sharding cuts latency 07:43 The hot shard problem 08:21 Resharding and the shard tree 09:01 Choosing a shard key Who this is for: backend engineers, system design interview prep, and anyone who has only ever run a single-instance database. #systemdesign #database #sharding #backend #scalability