When a range rowid filter was used with an index ref access the cost of
accessing the index entries for the records rejected by the filter was not
taken into account. For a ref access by an index with big average number
of records per key this led to poor execution plans if selectivity of the
used filter was high.
The patch resolves this problem. It also introduces a minor optimization
that skips look-ups into a filter that turns out to be empty.
With this patch the output of ANALYZE stmt reports the number of look-ups
into used rowid filters.
The patch also back-ports from 10.5 the code that properly sets the field
TABLE::file::table for opened temporary tables.
The test cases that were supposed to use rowid filters have been adjusted
in order to use similar execution plans after this fix.
Approved by Oleksandr Byelkin <sanja@mariadb.com>
ANALYZE and ANALYZE FORMAT=JSON structures are changed in the way that they
show additional information when rowid filter is used:
- r_selectivity_pct - the observed filter selectivity
- r_buffer_size - the size of the rowid filter container buffer
- r_filling_time_ms - how long it took to fill rowid filter container
New class Rowid_filter_tracker was added. This class is needed to collect data
about how rowid filter is executed.
A 64bit counter can overflow within the time of a query
so lets take it that the measurement is the small value
rather than an order 1e12 millisecond query.
tested with:
int main()
{
ulonglong start = ULONGLONG_MAX - 30;
ulonglong end = 600;
ulonglong cycles = 10000;
cycles += end - start;
if (unlikely(end < start))
cycles += ULONGLONG_MAX;
printf("cycles %llu\n", cycles);
}
- Remove ANALYZE's timing code off the the execution path of regular
SELECTs.
- Improve the tracker that tracks counts/execution times of SELECTs or
DML statements:
= regular execution just increments counters
= ANALYZE will also collect timings.
