Traces and topology

A trace answers what one request crossed. Topology answers what those requests have built. Both use the same data.

Traces

Distributed tracing and waterfalls
SQL, HTTP and messaging inside spans
Error attribution and stack traces
Two-way links to logs and sessions

Topology

Runtime map: services, instances and dependencies
Impact paths: which upstreams a slow request affects
Service flow: traffic and latency from entry to exit
Flame graphs: duration aggregated by call stack
Relationship history: when the graph changed

Where relationships come from

Relationships come from real calls, not hand-maintained configuration. Edges between services, hosts, containers, databases and queues are generated from captured data.

Suggested workflow

Use topology to scope the blast radius
Drill from a node into the exact trace
Confirm with logs and session replay

Limits

Estates with no east-west traffic and no network capture get an incomplete graph
Calls across NAT or network segments need address mapping checked
Relationship history depends on retention; older snapshots expire