Problems / Real world, hard

Real-time document editing: an all-hands document

Asked at Google (Docs) and Figma. Every keystroke is a transaction that locks the document row in Postgres, cursors are database writes, and opening a document replays it from the database. At 8 s a company all-hands starts and traffic triples for 10 seconds.

Unlock it once for 25 credits; everything inside it is free (runs, changes, Submit, results, hints). Sign in with Google to start; new accounts get free credits.

Your goals

Success ≥ 99.5%, p99 ≤ 400 ms and Apdex ≥ 0.90 through the meeting, for at most $13,000 a month.

Test cases

Like a coding problem, a solution has to pass every case, not just the one it was built for. Submit in the lab runs all 4; 1 of them are hidden until then.

  1. As given. The problem’s own traffic, scheduled failures and seed.
  2. A slow dependency. The main data store or model answers twice as slowly for part of the run.
  3. A server dies. One replica of the service with the most replicas is killed for part of the run.

The traffic

8,000 requests a second for 20 seconds: 60% edit (writes), 30% open (reads), 10% cursor (writes).

The system

What each component does in this design, and what calls it. The simulator mimics each one from these settings.

ComponentIts job hereSettings
Load balancerLoad balancerTraffic enters here.Calls Docs API.Layer 4 balancers route by IP and port; layer 7 balancers read HTTP and can route by path or header.–
Docs APIApp serviceLoad balancer forwards to it.Calls Documents and edits (Postgres, row lock per document).Little's law: throughput ≈ concurrency ÷ latency. 200 slots at 50 ms serve about 4,000 req/s.8 replicas × 100 slots · timeout 1.0 s
Documents and edits (Postgres, row lock per document)SQL databaseDocs API calls it.Postgres runs one process per connection, which is why poolers like PgBouncer sit in front of busy databases.16 vCPU · 400 connections

More real world problems