Why Pinterest Bet on Sharded MySQL Over NoSQL
Pinterest hit early scaling walls, tried NoSQL alternatives, then built ID-encoded sharded MySQL instead — and it outlasted the exotic options.
Engineering at scale
9 articles · 36 min of reading
Pinterest hit early scaling walls, tried NoSQL alternatives, then built ID-encoded sharded MySQL instead — and it outlasted the exotic options.
How Pinterest built shared machine learning feature infrastructure to rank home feed content consistently across training and real-time serving.
How Pinterest migrated thousands of services onto Kubernetes and built internal platform abstractions so product teams didn't need to become infrastructure experts.
Behind Pinterest's sharded MySQL fleet sits a heavy memcache layer, and the real engineering challenge was never caching reads — it was invalidating them correctly.
A look back at the years Pinterest ran core data on HBase, why the wide-column model fit some of its access patterns, and where the friction showed up.
How Pinterest built visual search and object detection to let people discover products and ideas by tapping on specific items inside an image.
Pinterest built Teletraan to let any engineer deploy their own service safely, with staged rollouts and fast rollback replacing manual, ops-gated pushes.
Pinterest's Pixie system serves billions of recommendations in real time by running random walks over an enormous pin-and-board graph held in memory.
How Pinterest replaced a fragile in-memory queue with PinLater, a durable, pluggable asynchronous job execution system built to survive real production load.