• full-stack javascript
• database sync across clients
• use the database query syntax directly
• client-side cache
• latency compensation (optimistic UI updates)
• RPC protocol

• Query the initial result into working set
• Tail MongoDB replication log
• Analyze every operation matching our set
• Refetch the document in question or apply the change
• Gracefully handle race conditions
• Refetch the whole query if the log is too far ahead
• Proxy/filter replication log in a separate process

Oplog strategy

filtering, ordering, limits, projections

almost every Mongo operator is supported

Meteor and MongoDB Oplog

http://youtu.be/_dzX_LEbZyI

Meteor and Redis

http://youtu.be/-Vnb8tjnE3k

• Miniredis
  • strings,lists, maps
• Key-Value notifications to get updates about changes
• No filtering or querying

Match your DB with Meteor!

• Build the realtime driver
• Build the front-end cache
• Glue them with DDP and RPCs
• ???
• PROFIT

... or teach it speak DDP

https://meteor.com