New Relic
NRQL & Dashboards
Query all telemetry with a SQL-like language and build custom dashboards.
NRQL — the New Relic Query Language — is the SQL-like language for everything in NRDB. You SELECT functions (count, average, percentile) FROM an event or metric type, filter with WHERE, group with FACET, and bucket over time with TIMESERIES. Because metrics, logs, traces, and custom events all live as queryable data, one language answers "error rate by service", "p95 latency faceted by endpoint", and "count of a custom business event" alike.
A single query desk at a vast archive: you write one standard request form and the desk fetches receipts, photos, or letters alike — then charts them however you ask.
Key Concepts
1
Dashboards are collections of NRQL-powered widgets. The same query language underlies alerts, so a chart you build can become an alert condition directly. NRQL also supports useful analytics primitives — funnels, histograms, percentiles, and the ability to query custom attributes you attach to events — which makes it a tool for product and business metrics, not just infrastructure. Mastering FACET and TIMESERIES is the practical core most questions probe.
2
What the interviewer is probing.
What the interviewer is probing.
3
1. "What is NRQL?" *Probing:* the basics. *Stalls:* "A language." *Moves up:* a SQL-like query
language over the telemetry database, with FACET for grouping and TIMESERIES for charting.
1. "What is NRQL?"
4
2. "What does FACET do?" *Probing:* the key operator. *Stalls:* "Filters." *Moves up:* groups
results by an attribute, which is how one query becomes a per-service or per-endpoint breakdown.
2. "What does FACET do?"
5
3. "How do you compare against last week?" *Probing:* the practical idiom. *Stalls:* "Two
queries." *Moves up:* the COMPARE WITH clause, which overlays the earlier period on the same chart.
3. "How do you compare against last week?"
6
4. "What is the performance consideration?" *Probing:* the practical angle. *Stalls:* "None."
*Moves up:* wide time ranges over high-cardinality facets are slow and costly — narrow the window
and the facet first.
4. "What is the performance consideration?"