Rick Copeland describes the results of his experiment designing and testing a MongoDB-based monitoring system. Some of his findings are quite interesting:
- So documents that grow and grow and grow are a real performance-killer with MongoDB.
- BSON actually stores documents as an association list thus lookups are not constant-speed
One aspect that the post doesn’t cover is how MongoDB behaves when there’s a lot of data contention which is usually the case for tracking systems.
Original title and link: Designing a MongoDB Schema for a Monitoring System ( ©myNoSQL)