Problem Statement
Sentry has spans and distributed tracing for a while now. It would be very useful to provide a way to see how each service (or project) are connected in a very big picture. One way of achieving that is through topology graph.
Solution Brainstorm
Jaeger has this:
https://www.jaegertracing.io/docs/2.16/features/#topology-graphs
https://demo.jaegertracing.io/jaeger/dependencies
A trace waterfall follows the path of a single request through your system — you can see each service it hit, in what order, and how long each hop took.
A topology graph zooms out: it aggregates across all your traces to build a map of your entire system, showing which services talk to each other and where the hottest paths (highest traffic or latency) are.
Think of it like: the waterfall is one car's GPS route, the topology graph is the traffic heat map of the whole city.
It gives you a high-level view of your system — the components that make it up, the hot paths between them, which parts are most error-prone, and how services depend on each other (so you can reason about blast radius when something goes wrong).
Signoz has a very similar feature: https://signoz.io/docs/userguide/service-map/
As well as Honeycomb: https://www.honeycomb.io/platform/service-map
Internal Slack thread: https://sentry.slack.com/archives/C08SBFDKBKJ/p1773347557420019
Product Area
Explore
Problem Statement
Sentry has spans and distributed tracing for a while now. It would be very useful to provide a way to see how each service (or project) are connected in a very big picture. One way of achieving that is through topology graph.
Solution Brainstorm
Jaeger has this:
https://www.jaegertracing.io/docs/2.16/features/#topology-graphs
https://demo.jaegertracing.io/jaeger/dependencies
A trace waterfall follows the path of a single request through your system — you can see each service it hit, in what order, and how long each hop took.
A topology graph zooms out: it aggregates across all your traces to build a map of your entire system, showing which services talk to each other and where the hottest paths (highest traffic or latency) are.
Think of it like: the waterfall is one car's GPS route, the topology graph is the traffic heat map of the whole city.
It gives you a high-level view of your system — the components that make it up, the hot paths between them, which parts are most error-prone, and how services depend on each other (so you can reason about blast radius when something goes wrong).
Signoz has a very similar feature: https://signoz.io/docs/userguide/service-map/
As well as Honeycomb: https://www.honeycomb.io/platform/service-map
Internal Slack thread: https://sentry.slack.com/archives/C08SBFDKBKJ/p1773347557420019
Product Area
Explore