Per‑operation view of the service’s database calls — find the slow or failing query behind a service’s latency, and trace it in context.

Get here: Service Overview → Databases tab.

What this panel tells you. It isolates the outbound calls this service makes to databases (transaction_category = Database), grouped by query/operation and DB system (db.system) — the panel behind the DB Operations card on Service Overview. Use it to separate a data‑layer problem from application code: if a service’s latency rose but its own web logic is fine, a query here is usually the cause. Most Time Consuming surfaces the query that drains the most total time (a fast query run millions of times can outrank one slow query), Highest Error Rate flags failing statements (timeouts, constraint/connection errors), and Highest Throughput highlights chatty queries that are prime candidates for caching or indexing.

Charts (trace metrics — exact)

Four trend charts — Latency, Error Rate (non‑web), Throughput, and Most Time Consuming — track how the service’s database calls behave over the selected window. Click a chart to drill in, brush across it to zoom, or expand it to full size.

Filters

Narrow the view by Infrastructure (cluster/namespace) and Resource (a specific pod/host, or All Resources), or focus on one Database sub‑category (e.g. by db.system). Metric controls let you set the Error Code series (5.*) and Apdex Threshold (500 ms), and the Compare / Range / Step options change the time window and comparison.

Top 20 Operations

Ranked three ways; a Sub‑category column groups by database system (db.system) and rows can show the DB query text. All Operations opens the full searchable list.

Most Time Consuming — which queries cost the most time?

  • Operation / Sub‑category — the query and DB system. Click to open the operation detail.
  • Slowest Span — slowest sampled query span (clickable); jump to a worst‑case example → waterfall.
  • Overall Latency (Ms) — cumulative time consumed (latency × volume); highest = biggest total drain even if each call is fast.
  • Peak Avg Latency (Ms) — worst spike of the average; big gap vs. Avg = bursty (contention/locks).
  • Avg Latency (Ms) — mean query latency (baseline).

Highest Error Rate — which queries fail the most?

Operation / Sub‑category · Error Rate (%) (non‑web error rate) — cross‑check throughput to prioritize.

Highest Throughput — which queries run most often?

Operation / Sub‑category · Throughput (RPM) — busiest queries; prime candidates for caching/indexing.

Operation detail

Click a query/operation → tiles + Segment Breakdown → View Traces → open a sampled trace → waterfall to confirm the DB span is the bottleneck.

Workflow

  1. Brush the Latency / Most Time Consuming chart to the problem window.
  2. Use Database sub‑category + Resource to isolate a system/pod.
  3. Open Most Time Consuming / Highest Error Rate → identify the query.
  4. Drill to the trace to see the query within the request.

Share permalink · Export CSV · Top 20 cap on ranked tables.