Monitor system performance

My monitoring dashboard shows CPU and memory spikes on web01 and db02 over the last 24 hours. Send the attached performance summary to Lina in SRE and Mark, the on-call DB lead, asking both to review and sign off by Friday so we can escalate to procurement for capacity if needed. Check the graphs for missing metrics and confirm collection intervals before sending.

Before I share this performance report with SRE and the product team, make the hot spots easy to act on: put the 95th-percentile CPU and memory numbers for web01 and db02 at the top, link each to the time range with the highest spike, and flag any missing metrics or collection gaps that would make the data unreliable for capacity decisions.

Database CPU jumped to 95% at 02:14 during backups and the alert page shows no corresponding IO or swap metrics. I told Lina and the DB lead but I am not sure if this is a noisy alert, a metric collection gap, or real contention. What’s the most likely cause and what should I check first to avoid a false escalation?

Every month we get paged for high CPU or memory on the same two hosts, spend nights adding temporary knobs, then forget to change retention or fix the root cause. Where are we wasting time and credibility in our monitoring runbook, and which single habit should we change so we stop firefighting the same hosts every month?