Per‑operation view of the service’s web (HTTP server) work — find the endpoint that is slow, failing, or driving the most traffic, then jump to a real trace.
Get here: Service Overview → Web Operations tab.
What this panel tells you
It covers the inbound HTTP/RPC requests the service answers (transaction_category = HTTP | RPC Systems, kind = server) — i.e. the service’s own endpoints, one row per operation (GET /orders, POST /checkout…). This is the panel behind the Web Operations card on Service Overview. Use it to answer which endpoint is the problem and why: a high error rate points at the endpoint’s own logic or a bad dependency, rising latency with flat throughput usually means a downstream call got slower, and a dominant throughput row is the busiest path (biggest blast radius). When an endpoint looks healthy here but the request is still slow end‑to‑end, the cost is downstream — pivot to Databases, Messaging, or External Services via its Segment Breakdown.
Overview banner
A strip of headline numbers for the current filter — Max Latency, Average Latency (Ms), Max Throughput (RPM), Highest Error Rate, and Total Count — giving you a one‑glance read on how the service’s web endpoints are doing before you dig into individual operations.
Service Metrics charts (trace metrics — exact)
Four trend charts — Latency (ms), Error Rate (%), Throughput (RPM), and Apdex Score — each plotting the overall total plus a separate line per top operation (GET, POST…), so you can see which endpoints move the needle. Click a chart to drill in, brush across it to zoom into a window, or expand it to full size.
Filters
- Data: Infrastructure (cluster/namespace) · Resource (pod/host, All Resources) · Sub‑category (e.g. request method).
- Metric: Error Code (Code Series,
5.*) · Apdex Threshold (500 ms). - Compare / Range / Step via the Compare With dropdown.
Top 20 Operations
Ranked three ways — each tab answers a different question. View All Operations opens the full, searchable list.
Most Time Consuming — which operations cost the most time?
- Operation — the endpoint/method. Click to open the operation detail (Segment Breakdown).
- Slowest Span — duration of the single slowest sampled span in the window (clickable); jump to a worst‑case exemplar → waterfall.
- Peak Avg Latency (Ms) — worst spike of the operation’s average latency; big gap vs. Avg = spiky, close = consistently slow.
- Avg Latency (Ms) — mean latency across the window (baseline; default sort).
Highest Error Rate — which operations fail the most?
- Operation — the endpoint/method. Click for the error breakdown.
- Error Rate (%) — % of executions that errored (per Error Code Series,
5.*); cross‑check throughput so a high‑error, high‑traffic op wins.
Highest Throughput — which operations carry the most traffic?
- Operation — the endpoint/method. Click for detail.
- Throughput (RPM) — requests per minute; busiest paths (biggest blast radius; capacity planning).
Operation detail (drill‑in)
Click an operation → tiles (P99/P95 Latency, Error Rate, Total Error Count, Throughput, Total Operations), Segments – Latency chart, and a Segment Breakdown table (Sub Category · Segment · Service · Avg Latency · Avg Throughput) that shows which downstream call dominates. Header: View Traces and Share.
See the full path in Service Overview → Drill‑down flow.
Share / Export / Limits
Share permalink · Export CSV of the current grid · ranked tables show Top 20 only.