Atlas and Developer Tooling
Atlas building blocks
Atlas is the managed database service. You create a project, a cluster (shared tiers like M0, or dedicated tiers M10+), database users, and a network access list. Database users are MongoDB credentials (SCRAM); the IP access list is a separate firewall mechanism - adding a user does not grant network access, and vice versa. Clusters expose an SRV connection string, support horizontal and vertical scaling, autoscaling on dedicated tiers, and continuous backups with point-in-time restore. Feature availability differs by tier, so verify the matrix rather than assuming a free-tier cluster behaves like a dedicated one.
Tools a developer is expected to know
mongosh is the modern shell (JavaScript REPL, .mongorc.js, db.getSiblingDB(), load()). Compass is the GUI for browsing documents, building queries and aggregation pipelines visually, and reading the explain plan. The Atlas UI adds Data Explorer, a real-time performance panel, the Performance Advisor (suggests indexes from slow queries), Atlas Search (Lucene-based full-text search with its own index definitions), Atlas Charts, Triggers, and Online Archive for tiering cold data. The Atlas CLI and Admin API automate cluster management, and the VS Code extension embeds a playground for running queries against a cluster.
Traps
Performance Advisor recommends indexes but you still validate with explain. Atlas Search indexes are separate objects from MongoDB indexes and are queried with the $search stage, which must be the first stage of a pipeline. mongosh commands such as show dbs are shell helpers, not driver API. Backup restore creates a new cluster or restores into an existing one, so plan for connection-string changes.
Worked example
A slow query on Atlas: check the Performance Panel for the operation, open Performance Advisor for a suggested index, create it in the target environment via mongosh or Compass, then confirm with explain("executionStats") that IXSCAN replaced COLLSCAN.