
If Your Version is 11.5 or 11.9.x
54
Adaptive Server can capture query text and save an abstract plan for a query in
a new system table called
sysqueryplans. Using a rapid hashing method,
incoming SQL queries can be compared to stored query text, and if a match is
found, the saved abstract plan is used to execute the query.
For more information on abstract query plans, see the
Performance and Tuning
Guide at http://manuals.sybase.com:80/onlinebooks/group-
as/asg1250e/perf/@Generic__BookView
. For a quick overview of how to use
abstract query plans to troubleshoot performance problems after upgrade, see
the TechNote
Using and Maintaining Abstract Query Plans at
http://my.sybase.com/detail?id=1010648
Increased Optimization Time
Queries with long chains of join keys may require additional time to optimize
with Adaptive Server. If the time required to optimize such a query is
unacceptable, consider using an abstract query plan for the query.
Like Optimization Enhancements
ASE 12.0 changes the costing for like clauses that are not migrated into
SARGs, using a technique of generating more accurate cost estimates for
queries that include a leading wildcard in a
like clause.
This provides better selectivity estimates, resulting in better query plans.
Previously a like clause with a leading wild card was estimated to qualify all
rows in the column (a selectivity of 1.0) since there was no way to search for a
match. This was not the most accurate way to cost such a clause.
Example:
select ... from part, partsupp, lineitem
where l_partkey = p_partkey
and l_partkey = ps_partkey
and p_title = ’%Topographic%’
The like string is compared with histogram cell boundaries.
A match occurs when the like term is found in a cell boundary by means of a
pattern match search.
If no pattern match is found the selectivity is assumed to be 1/number of steps
in the histogram. If the default of 20 cells is used, selectivity will be 0.05 if
there is now pattern match in the histogram boundary values.
Comentários a estes Manuais