5 exercises — in incident updates, post-mortems, and quarterly reviews, precise trend vocabulary replaces vague phrases like "it was high" or "it changed".
Spike:spiked briefly (sharp rise + recovery); surged and remained elevated (no recovery)
Stable:stable / flat / holding steady / fluctuating within a narrow range
Continuous growth:growing continuously / climbing without recovery → memory leak signal
0 / 5 completed
1 / 5
A Grafana dashboard shows CPU utilisation going from 22% to 91% in under two minutes, starting at 03:14 UTC during a production deployment. How should an on-call engineer describe this in an incident update?
"Rose sharply" is the correct phrasing for a rapid, steep increase. Key vocabulary for upward trends:
• rose / increased / climbed sharply / steeply / rapidly — fast upward movement • rose gradually / slowly / steadily — slow, consistent upward movement • surged — sudden dramatic spike (often implies a problem) • peaked at X% — reached the highest point
Distractors: "decreased gradually" is the opposite. "Leveled off" means it stabilised (flat line). "Fluctuated" means it went up and down repeatedly — not a single spike.
In incident write-ups: precise direction + speed vocabulary (rose sharply, dropped suddenly, spiked briefly, climbed steadily) makes timeline reconstruction much faster in post-mortems. Vague language like "CPU was high" loses the timing and rate information that is critical for root cause analysis.
2 / 5
A monitoring chart shows the error rate at 0.08% for 6 hours. At 14:32, it jumps suddenly to 5.4%, then returns to 0.09% within 4 minutes — coinciding with a canary deployment that was automatically rolled back. Which description is most accurate?
"Spiked briefly before returning to baseline" captures both elements: sudden high peak + rapid return to normal. Key vocabulary:
• spiked — sudden sharp increase and return (like a spike on a graph); implies transient • briefly — for a short time (4 minutes) • returned to baseline — went back to the normal / expected level • baseline — the normal/expected value under typical conditions
Distractors: "surged and remained elevated" implies it stayed high — not true here (it came back down). "dropped sharply" is the wrong direction. "gradually increased" is wrong direction and wrong speed.
More precise alternatives: "The error rate spiked to 5.4% at 14:32 for approximately 4 minutes before the canary rollback restored it to baseline." This is the standard post-mortem phrasing: spike time, peak value, duration, and recovery mechanism.
3 / 5
A performance review shows API response latency (p99) over the past two weeks: the numbers are 448ms, 451ms, 452ms, 449ms, 453ms — varying by a few milliseconds each day with no clear trend. How would you describe this trend in a weekly report?
"Broadly stable / fluctuating within a narrow range" — both parts of this description are accurate and precise. Vocabulary for flat / stable trends:
• stable / broadly stable — no meaningful change • flat — completely unchanged (or nearly so) • fluctuating within a narrow range — small variations but no trend • holding steady at X — staying at approximately the same value • showing no significant trend — statistical language for no upward or downward movement
Distractors: "increased dramatically" and "degrading rapidly" are dangerous overstatements — a 5ms variation on 450ms baseline is normal noise, not a trend. In SRE communication, precision matters enormously: falsely alerting the team to a "dramatic increase" erodes trust and causes alert fatigue. "Dropped significantly" is factually wrong.
4 / 5
Active user data over four months: January: 10,200 DAU — February: 11,800 DAU — March: 13,400 DAU — April: 15,100 DAU. How would you describe this to the product team in a quarterly review?
"Growing steadily — up approximately 48% over four months" — the best answer combines direction vocabulary with a quantified magnitude. Key vocabulary for consistent upward trends:
• growing steadily / consistently — continuous, reliable growth • increasing month over month (MoM) — growing each consecutive period • up X% over the period — quantified with percentage change • trending upward — clear positive direction • showing healthy growth — positive framing for stakector reports
Distractors: "fluctuating wildly" — wrong, the numbers go up consistently. "dropped sharply" — factually wrong. "peaked and plateau" — speculation, not supported by the data given.
Best practice: Always include both the qualitative trend description ("growing steadily") and the quantitative measure ("up 48%") for maximum clarity in business communication.
5 / 5
A DevOps engineer monitors heap memory for a service that has been running for 24 hours without a restart: at 00:00 it was 1.8 GB, at 06:00 — 2.9 GB, at 12:00 — 4.1 GB, at 18:00 — 5.4 GB, at 24:00 — 6.8 GB. How should the engineer describe this pattern in a bug report?
"Growing continuously — consistent with a memory leak" — the trend is a classic linear/monotonic increase with no recovery, which is the hallmark of a memory leak. Key vocabulary:
• growing continuously / climbing steadily without recovery — the defining pattern of a leak • memory leak — a defect where memory is allocated but never released, causing progressive memory exhaustion • unbounded growth — a system metric that increases without an upper limit • OOM (Out of Memory) — the eventual crash when heap is exhausted • heap exhaustion — when the JVM/process runs out of heap memory
Contrast with normal patterns: • Saw-tooth pattern: memory rises then drops sharply → normal GC cycles • Spike then recovery: temporary high memory under load that returns to baseline → normal • Step increase then stable: one-time increase + stable → possibly a cache fill, not a leak
The pattern here (1.8 → 2.9 → 4.1 → 5.4 → 6.8 GB, +1–1.3 GB every 6h) = linear growth with no recovery = almost certainly a memory leak.