www.infinidb.org

I installed 1.0.2 and during the install it created a Calpont folder in usr/local. The queries ran well, and a major query in my system loaded in just under 2 sec.

I made a backup of this as Calpont-2.

I then went through a few passes at installing 1.0.3 then back to 1.0.2.

Now the speed that I am getting is slower. The exact same query is taking over 7 sec.

If I backup Calpont folder, and re-instate Calpont-2 as Calpont, then I get the good speed again.

But why could I be getting a better speed in this case? The query is same and the version is same (1.0.2). I want to understand so that i can install 1.0.3 and get the same good speed.

Any ideas how to check?

Hey Rasel, let see if we can figure out what is going on.

Can you post the following info and we can see diagnose in-house:

1. The create table command(s)
2. How many rows are being insert into the table.
3. Small snipet of the import file
4. The query being run

Thanks, David

Rasel,

We investigated and the extent map was not populated for all of the columns involved in the where clause for the query. The optimizer was choosing a less efficient plan in the new instance as it did not have all of the extent map information available to choose the optimal plan. We have the feature to auto populate the extent map in our roadmap that can be seen here - www.infinidb.org/resources/tech-articles...nity-edition-roadmap .

In the current release, a column's extent map is populated by running a query that scans a column. For example, "select count(*) from account where accountID < 0;" will populate the extent map for the accountID column. The Performance and Tuning Guide available at www.infinidb.org/downloads/cat_view/71-a...d-user-documentation discusses this in further detail.

Thanks,
Walt

Thanks Infinidb. Now i understand clearly about extent map. and my query runs with nice timing.
Thanks again for your help!!