I Built Consistent Hashing From Scratch in Go — Here's What I Learned

If you've ever added a server to a cache cluster and watched your database melt, you already know the problem consistent hashing solves. You just might not know it by name. I built a full implementation from scratch in Go to understand it deeply. This post walks through what I learned — the problem, the fix, and the gotchas nobody tells you about. The five-minute version You have 5 cache servers. You route keys with hash(key) % 5. Life is good. Then traffic spikes and you add a 6th server. Now it's hash(key) % 6. Sounds harmless, right? Here's what actually happens: Before: hash("user:1001") % 5 = 3 → Server C After: hash("user:1001") % 6 = 1 → Server A ← moved! That key was sitting happily on Server C. Now every client thinks it's on Server A, where it doesn't exist. Cache miss. The request hits your database. And it's not just one key. ~83% of all keys remap. With 100K cached items, that's 83,000 simultaneous cache misses hammering your database. People call this a "thundering herd"