Вдогонку к своему-же письму. E:\work\tmp\perf_sdb>perf_sdb_test.exe --conn_str="/InterBase_EMB" -- db_name=TES T01.FDB --user_name=test --user_passwd=test
Sdb performance tests ... Test INSERT 01. Insert 4000 records. Autocommit. INSERT INTO t1 VALUES(:?, :?, :?) 0.062 sec. Test INSERT 02. Insert 100000 records. Inside of a transaction. INSERT INTO t2 VALUES(:?, :?, :?) 2.109 sec. Test INSERT 03. Insert 100000 records into an indexed table. Inside of a transac tion. INSERT INTO t3 VALUES(:?, :?, :?) 26.907 sec. Test SELECT 02. Select 100 times without an index. Fetch result. SELECT count(*), avg(b) FROM t2 WHERE b >= :? AND b < :? 0.093 sec. Test SELECT 03. Select 100 times without an index on a string comparison. Fetch result. SELECT count(*), avg(b) FROM t2 WHERE c LIKE :? 0.157 sec. Test INNER JOIN 01. Without an index. Fetch result. SELECT t1.a FROM t1 INNER JOIN t2 ON t1.b = t2.b 0.218 sec. Test CREATE INDEX 01. Run 1 time on an ordered field. CREATE INDEX i2a ON t2(a) 0.266 sec. Test CREATE INDEX 02. Run 1 time on a nonordered field. CREATE INDEX i2b ON t2(b) 0.266 sec. Test SELECT 04. Select 1000 times with an index. Fetch result. SELECT count(*), avg(b) FROM t2 WHERE b >= :? AND b < :? 0.015 sec. Test UPDATE 01. Update 400 times without an index. Inside of a transaction. UPDATE t1 SET b = b*2 WHERE a >= :? AND a < :? 1.704 sec. Test UPDATE 02. Update 4000 times with an index. Inside of a transaction. UPDATE t2 SET b = :? WHERE a = :? 0.562 sec. Test UPDATE 03. Update 4000 times a text value with an index. Inside of a transa ction. UPDATE t2 SET c = :? WHERE a = :? 0.188 sec. Test INSERT from a SELECT 01. Inside of a transaction. INSERT INTO t1 SELECT * FROM t2 0.656 sec. Test INSERT from a SELECT 02. Inside of a transaction. INSERT INTO t2 SELECT * FROM t1 21.547 sec. Test INNER JOIN 02. With index on one side. Fetch result. SELECT t1.a FROM t1 INNER JOIN t2 ON t1.b = t2.b 2.953 sec. Test SELECT 05. Select 100 times with subqueries. Subquery is using an index. Fe tch result. SELECT t1.a FROM t1 WHERE t1.b IN (SELECT t2.b FROM t2 WHERE t2.b >= :? AND t2.b < :?) 2.062 sec. Test DELETE 01. Run 1 time without an index. DELETE FROM t2 WHERE c LIKE '%fifty%' 0.25 sec. Test DELETE 02. Run 1 time with an index. DELETE FROM t2 WHERE a > 10 AND a < 95000 1.688 sec. Test DELETE 03. A big DELETE followed by many small INSERTs. Run 4000 times. Inside of a transaction. DELETE FROM t1 INSERT INTO t1 VALUES(:?, :?, :?) 0.984 sec. Для сравнения тесты для SQLITE. E:\work\tmp\perf_sdb>perf_sdb_test.exe --conn_str="/SQLITE_EMB" -- db_name=test. sqlite Sdb performance tests ... Test INSERT 01. Insert 4000 records. Autocommit. INSERT INTO t1 VALUES(:?, :?, :?) 0.016 sec. Test INSERT 02. Insert 100000 records. Inside of a transaction. INSERT INTO t2 VALUES(:?, :?, :?) 0.641 sec. Test INSERT 03. Insert 100000 records into an indexed table. Inside of a transac tion. INSERT INTO t3 VALUES(:?, :?, :?) 2.406 sec. Test SELECT 02. Select 100 times without an index. Fetch result. SELECT count(*), avg(b) FROM t2 WHERE b >= :? AND b < :? 6.078 sec. Test SELECT 03. Select 100 times without an index on a string comparison. Fetch result. SELECT count(*), avg(b) FROM t2 WHERE c LIKE :? 9.781 sec. Test INNER JOIN 01. Without an index. Fetch result. SELECT t1.a FROM t1 INNER JOIN t2 ON t1.b = t2.b 159.8 sec. Test CREATE INDEX 01. Run 1 time on an ordered field. CREATE INDEX i2a ON t2(a) 0.937 sec. Test CREATE INDEX 02. Run 1 time on a nonordered field. CREATE INDEX i2b ON t2(b) 0.922 sec. Test SELECT 04. Select 1000 times with an index. Fetch result. SELECT count(*), avg(b) FROM t2 WHERE b >= :? AND b < :? 0.703 sec. Test UPDATE 01. Update 400 times without an index. Inside of a transaction. UPDATE t1 SET b = b*2 WHERE a >= :? AND a < :? 0.563 sec. Test UPDATE 02. Update 4000 times with an index. Inside of a transaction. UPDATE t2 SET b = :? WHERE a = :? 0.109 sec. Test UPDATE 03. Update 4000 times a text value with an index. Inside of a transa ction. UPDATE t2 SET c = :? WHERE a = :? 0.078 sec. Test INSERT from a SELECT 01. Inside of a transaction. INSERT INTO t1 SELECT * FROM t2 1.016 sec. Test INSERT from a SELECT 02. Inside of a transaction. INSERT INTO t2 SELECT * FROM t1 2.875 sec. Test INNER JOIN 02. With index on one side. Fetch result. SELECT t1.a FROM t1 INNER JOIN t2 ON t1.b = t2.b 0.484 sec. Test SELECT 05. Select 100 times with subqueries. Subquery is using an index. Fe tch result. SELECT t1.a FROM t1 WHERE t1.b IN (SELECT t2.b FROM t2 WHERE t2.b >= :? AND t2.b < :?) 21.391 sec. Test DELETE 01. Run 1 time without an index. DELETE FROM t2 WHERE c LIKE '%fifty%' 0.141 sec. Test DELETE 02. Run 1 time with an index. DELETE FROM t2 WHERE a > 10 AND a < 95000 1.828 sec. Test DELETE 03. A big DELETE followed by many small INSERTs. Run 4000 times. Inside of a transaction. DELETE FROM t1 INSERT INTO t1 VALUES(:?, :?, :?) 0.093 sec. При работе с SQLITE процесс занимает порядка 3 Mb, при работе с Embedded Firebird процесс занимает порядка 30 мег. On Apr 10, 3:04 pm, "Novo" <[EMAIL PROTECTED]> wrote: > У меня есть тесты, сделанные по образцу SQLite. > Сделаны они были для тестирования различных API. > Написано все на C++.
