Stripe architecture: how Stripe scaled (3 case studies)
Stripe's data runs on document databases: MongoDB in its 2017 post and DocDB, its own document database platform, in 2024 (1.5 PB, 99.999% uptime). Stripe published a four-step dual-write playbook for migrating data models without downtime, later built a Data Movement Platform that moves data between shards without downtime, and protects its API with four kinds of rate limiters and load shedders built on Redis.
We track 3 public engineering write-ups by Stripe, published between 2017 and 2024. They cover dual writes, load shedding, online migration and rate limiting. Stack mentioned most: MongoDB (2). 1 of the 3 state concrete scale numbers, for example "1.5 PB". Each card below links to our short summary and the original post.
Tech used in Stripe's scaling write-ups
Stripe scaling timeline: every story, summarized
- 2017-02-02
Online migrations at scale
Stripe's four-step dual-write playbook for migrating data models without downtime.
- 2017-03-30
Scaling your API with rate limiters
The four kinds of rate limiters and load shedders Stripe uses to protect its API.
- 2024-06-06
How Stripe's document databases supported 99.999% uptime with zero-downtime data migrations
Stripe's DocDB and its Data Movement Platform move data between shards without downtime.
FAQ
What database does Stripe use?
Document databases: MongoDB in Stripe's 2017 online migrations post, and DocDB in its 2024 post, which cites 1.5 PB of data and 99.999% uptime.
How does Stripe do zero-downtime migrations?
Stripe's 2017 post lays out a four-step dual-write playbook for changing data models without downtime. Its 2024 post describes a Data Movement Platform that moves data between DocDB shards without downtime.
How does Stripe rate limit its API?
Stripe's 2017 post describes the four kinds of rate limiters and load shedders it uses to protect its API, built on Redis.