Subject | FB1.5.5 to FB2.1.3 migration |
---|---|
Author | SP |
Post date | 2009-12-01T22:58:36Z |
Hi,
We are planning to migrate from FB1.5.5 to FB2.1.3 (32bit) and then to
the 64 bit next.
While benchmarking the performance, we discovered that DB2.1.3 has
performed much slower than FB1.5.5.
I used a dummy database, created with FB1.5.5's QLI command (ODS 10.1),
create database test.fdb
create table test_table (numkey int, strvar varchar(25))
create unique ascending index numkey on test_table(numkey)
quit
I then ran the insertion and selection a few times on both FB1.5.5 and
FB2.1.3 installations (on same hardware, one after another), and timed
the QLI executions,
Insertion:
declare cnt long
cnt = 0
ready test.fdb
repeat 100000 begin
cnt = cnt + 1
insert into test_table (numkey, strvar) values (cnt, 'test')
end
commit
finish
quit
Selection:
ready test.fdb
repeat 10 select * from test_table
finish
quit
The performance in FB2.1.3 has found to be double the execution time of
that collected in FB1.5.5 for insertion and about 20% more execution
time than that collected in FB1.5.5 for selection.
Using ODS11.1 in FB2.1.3 improves a little but not much.
Is my observation correct?
Regards, SP
[Non-text portions of this message have been removed]
We are planning to migrate from FB1.5.5 to FB2.1.3 (32bit) and then to
the 64 bit next.
While benchmarking the performance, we discovered that DB2.1.3 has
performed much slower than FB1.5.5.
I used a dummy database, created with FB1.5.5's QLI command (ODS 10.1),
create database test.fdb
create table test_table (numkey int, strvar varchar(25))
create unique ascending index numkey on test_table(numkey)
quit
I then ran the insertion and selection a few times on both FB1.5.5 and
FB2.1.3 installations (on same hardware, one after another), and timed
the QLI executions,
Insertion:
declare cnt long
cnt = 0
ready test.fdb
repeat 100000 begin
cnt = cnt + 1
insert into test_table (numkey, strvar) values (cnt, 'test')
end
commit
finish
quit
Selection:
ready test.fdb
repeat 10 select * from test_table
finish
quit
The performance in FB2.1.3 has found to be double the execution time of
that collected in FB1.5.5 for insertion and about 20% more execution
time than that collected in FB1.5.5 for selection.
Using ODS11.1 in FB2.1.3 improves a little but not much.
Is my observation correct?
Regards, SP
[Non-text portions of this message have been removed]