Consistent Hashing: Adding Nodes Without Losing the Cache
Published on
Add one server to a cache cluster and the usual approach moves 89% of the cached data. Consistent hashing moves 12%. How it works, built and measured.
Read more →Published on
Add one server to a cache cluster and the usual approach moves 89% of the cached data. Consistent hashing moves 12%. How it works, built and measured.
Read more →Published on
Skip slow lookups for things that were never stored: a Bloom filter answers "definitely not here" for a million keys in 1.14 MiB instead of 88 MB.
Read more →Published on
When a cache fills up, something has to go. An LRU cache drops whatever was used longest ago, instantly. How to build one, and when to pick another rule.
Read more →
If you've journeyed this deep into my site, you're exactly who I want to connect with — whether it's about a fresh project or just a friendly chat. Feel free to reach out through social media or my contact page.