Rol

Jason Punyon

Data Science

Providence

Providence

Read API

Control how we use your data at https://stackoverflow.com/users/prediction-data

~300 Req/s at Peak

99th Percentile Latency ~3-5ms

SLA == 20ms

How do we do it?

(basically)

How do we do it?

Badly. 😭

What's the Problem?

What's the Problem?

What's the Problem?

What's the Problem?

What's the Problem?

So to sum up...

So to sum up...in GIF form

And what did I do about it?

...for months

So I had these problems floating around in my head for awhile, and this is what I made.
Interfaces as schema.
Consistency? LOL. We run most of Stack Overflow at Read Uncommitted.