Collections & indexes

Collections

The Collections section reuses the heat grid with a metric toggle — ops, latency or size — over a per-collection stats table underneath. The three views answer three different questions from the same $collStats data:

• ops — what is being hit;
• latency — what is slow (the grid switches to a warm ramp for this, because slow is not a “more is better” quantity);
• size — where the storage went.

The table below carries the underlying numbers: document count, storage size, average object size, index size.

Indexes

$indexStats reports accesses per index since the server started. That is a genuinely different thing from an index advisor’s opinion: an index with zero accesses has not been used, full stop.

So the Indexes section is a usage list sorted least-used first, with unused indexes flagged. Every index costs write throughput and storage, and the ones nobody is using are usually invisible — this is the screen that makes them visible.

Read it alongside Atlas’ suggested indexes if you run both: Atlas tells you which index to add, $indexStats tells you which to consider dropping.