Lab 1 (Standard HDD data disks): runs A to E and repeats A' to D' Redo measured on node-4 (last secondary of the distributed AG). Times in UTC, 2026-11-09/10. Runs F and G (a third mode with two undocumented trace flags) are left out; they behaved like parallel redo. Run log (workload start and end; the marker is the insert that closed the quiet period): run mode workload sessions duration_s start_utc end_utc marker_utc A serial bulk 40 0 2026-11-09T18:11:25Z 2026-11-09T18:12:44Z 2026-11-09T18:30:18.447 B parallel bulk 40 0 2026-11-09T18:46:28Z 2026-11-09T18:47:25Z 2026-11-09T19:05:24.994 C serial atomic 20 120 2026-11-09T19:22:37Z 2026-11-09T19:24:37Z 2026-11-09T19:42:47.507 D parallel atomic 20 120 2026-11-09T19:59:58Z 2026-11-09T20:01:59Z 2026-11-09T20:20:10.485 E1 parallel atomic 120 180 2026-11-09T20:29:26Z 2026-11-09T20:32:28Z 2026-11-09T20:49:37.783 E2 serial atomic 120 180 2026-11-09T21:07:16Z 2026-11-09T21:10:17Z 2026-11-09T21:27:30.582 Ar serial bulk 40 0 2026-11-09T21:42:33Z 2026-11-09T21:43:27Z 2026-11-09T21:47:34.232 Br parallel bulk 40 0 2026-11-09T22:04:42Z 2026-11-09T22:05:37Z 2026-11-09T22:09:36.366 Cr serial atomic 20 120 2026-11-09T22:26:43Z 2026-11-09T22:28:43Z 2026-11-09T22:32:56.989 Dr parallel atomic 20 120 2026-11-09T22:50:00Z 2026-11-09T22:52:01Z 2026-11-09T22:56:11.353 Log volume (msdb backup history on node-1) and when redo finished on node-4: run mode workload log_mb workload_s log_mb_s redo_finished_after_workload_end_s A serial bulk x40 1432.1 79 18 0.4 B parallel bulk x40 1433.2 57 25 1.1 C serial atomic x20 173.0 120 1.4 32.7 (outlier; repeat C' was 1.0) D parallel atomic x20 146.6 121 1.2 0 E1 parallel atomic x120 1234.8 182 6.8 0 E2 serial atomic x120 1284.0 181 7.1 0.4 A' serial bulk x40 1453.3 54 27 0.9 B' parallel bulk x40 1434.0 55 26 3.9 C' serial atomic x20 208.2 120 1.7 1.0 D' parallel atomic x20 176.2 121 1.5 0 Reading: at 27 MB/s of log or less, redo finished within 0 to 4 seconds of the workload in every mode. Redo was never the bottleneck, so these runs can't separate the modes.