Re: [Info-Ingres] Generate 1 million rows of unique data - Fridayfun

Thanks for all replies.

I used Kristoff's nojournaling suggestion.
3 seconds on test linux
11 seconds on the production Solaris boxes.
(see machine specs in an earlier mail)

I'm very happy now that my test server has topped MSSQL running on really
big machines.
But now I am on a mission.

Can someone say why the transaction log fills with over 300Mb of stuff?
And why these two cases perform so differently?

create table millions ... with nojournaling ; \g
insert into millions select .... ; \g
240 seconds

and

create table millions as
select .... with nojournaling ; \g
11 seconds

They are run in the same unjournalled uncheckpointed database. As far as I
can tell, the second statement does not hit the transaction log. Does the
nojournaling clause behave as nologging in this case?

Also, what sort of performance should I be expecting from our production
V880s. I think I may have some configuration work to do. It as if the query
is CPU bound.


FYI, the QEP top level node is the same for both queries.
Cart-Prod
Heap
Pages 12988 Tups 1000000
D25992 C22223

regards

Paul