🔗 The Hard Things About Sync

Sync engines for the web have gained popularity in recent years. Out of curiosity, I explored this space and was amazed to discover how many companies now work on open-source sync. Electric, Zero, Instant, Triplit, Supabase, RxDB, Convex are just some examples. Some provide libraries/frameworks that you can use in your own software stack, some offer a sync engine on top of your existing Postgres database, while others are taking another stab at building a modern realtime database for the web. Then there are companies like Figma, Notion and Linear that have implemented their own sync engines. Should you build or buy, or do you even need a sync engine at all? While there's no universal answer, I want to share my learnings from building a sync engine at Figma to help make sense of this technology, highlight some technical challenges of building your own, and suggest areas to consider when evaluating existing solutions.