Monitor Live Data Query Consumption

You can use the

Platform Consumption Console

to monitor consumption from your Live Data Query usage.

Rows Scanned

The

Rows Scanned

metric counts the rows scanned to run a query, not just the rows returned by the query.

You can try the following query patterns to reduce the number of rows scanned. Note that none of these patterns guarantee a reduction in rows scanned:

COUNT with no filter:
Using a count with no filter or something like

COUNT(id)

results in 1 row scanned.

COUNT with filters:
Rows scanned equals the number of rows after the filter is applied.

LIMIT:
In most cases, limits are pushed down. Limits up to 1000 scan the specified number of rows; limits above 1000 scan in pages of 1000 (Example: A

LIMIT 1001

query scans 2000 rows, i.e. 2 pages of 1000).

LIMIT 1

on a non-selective query scans ~1 row per split rather than the full table. Limit pushdowns don't apply to dynamic tables, join-pushdown tables, or

ORDER BY ... LIMIT (Top-N)

Examples

Assume a worker table of 100,000:

SQL	Rows scanned	Rows returned	Explanation
SELECT COUNT(*) FROM worker	1	1	Unfiltered count is answered without reading the table.
SELECT COUNT(*) FROM worker where name like 'j%'	5000	1	Filter is pushed down; 5000 rows match.
SELECT * FROM worker	100,000	100,000	Full read.
SELECT * FROM worker LIMIT 10	10	10	Limit is pushed down.
SELECT * FROM worker LIMIT 1000	1000	1000	Limit is pushed down.
SELECT * FROM worker LIMIT 1500	2000	1500	Rows are scanned in pages of 1000 so limiting by 1500 is 2 pages of 1000.
SELECT * FROM worker LIMIT 2000	2000	2000	Rows are scanned in pages of 1000 so limiting by 2000 is 2 pages of 1000.
SELECT * FROM worker WHERE date > 2001	40,000	40,000	Filter is pushed down; 40,000 rows match.
SELECT * FROM worker WHERE name LIKE 'j%'	40,000	40,000	Filter is pushed down; 40,000 rows match.
SELECT * FROM worker ORDER BY name LIMIT 2000	100,000	2,000	Limit pushdowns don't apply to ORDER BY ... LIMIT (Top-N) queries.

About Platform Consumption Console