Hint 1 — locked

Hint 2 — locked

Hint 3 — locked

Problem

Build a distributed counter service for tracking high-frequency events: YouTube view counts, Reddit upvotes, Instagram likes. Individual counters can receive millions of increments per second during viral events. The displayed count can be approximate but must eventually converge to the exact value.

Functional Requirements

• Increment a named counter atomically (e.g. post_123_likes)
• Read current counter value with < 50 ms latency
• Handle burst traffic (viral events: 1M increments/sec on a single counter)
• Persist counts durably — no loss on node failure
• Support counter reset and bulk read of multiple counters

Non-Functional

• 1M increments/sec per hot counter during viral events
• Displayed value can lag by up to 5 seconds (eventual consistency OK)
• 99.99% write availability — increments must never fail
• Exact count persisted eventually — no lost increments

Prerequisites