Drupal Migration — Performance Monitor

Bottom line — Spectre_5 is healthy on Drupal 11: the last 24 hours run 530–660ms response with apdex in the low-0.7s during peak traffic, and the week (Sep 23–29) is stable at 530–630ms with error rates under 0.05%. Since the Aug 1 upgrade, D11's phase aggregate (593ms avg, 0.23% errors, apdex 0.77 over 8.51M requests) is now the best-performing version on response time and error rate of all four phases tracked, edging out the D10 baseline (663ms, 0.36% errors) it replaced and the original D8 baseline (679ms, 0.39% errors) by a wider margin.

• D11 early read is already a solid sample, not a thin one — at ~8.51M requests over roughly two months post-upgrade, D11 has already out-volumed both D9 (4.50M) and D10 (3.11M) phase totals, so its improved numbers are not noise from a small window.
• Response time improved step-wise with the upgrade: D8 679ms → D9 699ms → D10 663ms → D11 593ms, the lowest average of the four phases.
• Error rate also improved: D8 0.39% → D9 0.42% → D10 0.36% → D11 0.23%, again the best of the four.
• One overnight latency/apdex dip in the last 24h: the 02:09 Chicago hour spiked to 1728ms average response with apdex dropping to 0.47, on very low traffic (1449 requests) — consistent with a single slow off-peak transaction skewing a thin bucket rather than a systemic issue, with a smaller echo at 06:09 (953ms). A brief error-rate blip (0.43%) also appears at 12:09 in an otherwise near-zero error trend.
• The Aug 1 05:00–07:00 Chicago maintenance window is excluded from the D10/D11 phase aggregates (D10 ends 05:00, D11 starts 07:00) so neither version is penalized for the deploy downtime, but that window still shows up as a visible traffic/response dip in the 24h and 7-day time series for anyone scrolling back to early August.

Method & caveats

• Cutovers. Drupal 9 / PHP 8 went live May 23, 2026; Drupal 10 Jul 4, 2026 — marked on every chart that spans them.
• Drupal 10 is an early read (~5 days). Its phase card uses the Jul 4–9 window directly.
• Recent spike. Jul 8 shows an isolated response jump (≈2× the prior day), coinciding with elevated reports/progress-API latency that day — an operational blip, not a migration regression.
• Source & retention. New Relic timeslice metrics (HttpDispatcher, Apdex, Errors/all), ~13-month retention — the full arc back to mid-2025 is covered. Percentiles (p95/p99) live in 8-day event data and are reported in the point-in-time reviews, not here.
• Refresh. Regenerated daily by a scheduled routine that re-queries New Relic and redeploys to this same URL.