Three steps: connect, choose what to compare, and copy your command. Everything else has a sensible default — open Fine-tune if you want to adjust it; the command on the right updates live as you go.
1 ·Connect
Any SQL*Plus logon:
user/pw@host:port/service, a TNS alias, or
/ as sysdba. Used locally — the password never leaves this page.2 ·What do you want to compare?
Pick the scenario closest to what you need — it fills in the windows, cadence and
template in Fine-tune below. ↺ Reset to defaults
AUTO ends the newest window at the most recent completed
clock hour. Pick a specific time to pin it — e.g. a Monday 09:00 or the end of a batch run.
3 ·Fine-tuneoptional ▸
Windows & cadence
How wide each compared window is, how many you compare, and the gap between them.
Length of every window. Below 1 reads as minutes (
0.25 = 15 min).How many earlier windows to compare against — a count (the name is historical).
The report needs ≥ 3 to compute z-score findings.
The gap between adjacent windows.
1 week (default) is the classic
same-hour-of-week trend; 1 hour walks straight back; 2 days is every
other day.Content & scope
How much detail to show and which instances to read.
Rows per ranking in Top SQL and the wait tables.
0 aggregates across all RAC instances; any other number filters
to that one instance.Timestamped one-liner per section while it runs — handy on a busy DB.
The HTML report is byte-identical whether this is on or off.
Timeline markersoptional ▸
Annotate the dated charts with milestones — a release, a patch, an incident.
They are baked straight into the command (no separate file to save), so it stays a single copy-paste.
Each milestone needs a date & time and a label. They ride inline in the command
(the
MARKERS env var for the shell, DEFINE markers for SQL*Plus).
A straight apostrophe in a label is swapped to a typographic one (’) so the command stays one
copy-paste; for exotic labels use a marker_file on disk instead.