OBIEE 11G - Oracle BI Server Cache (For seeding cache)

BenFreez

Hi All,

Here the situation, i have a report with more than 500 000 rows in the resultset and the query is taking 24 min to be executed.

The idea is to run the query during the night by using an ibot with the destination : Oracle BI Server Cache  (For seeding cache)

The ibot is executed but the resulset of the query is not in the cache.

I tried to increase in Enterprise Manager Maximum cache entry size and Maximum cache entries but the result is the same nothing in the cache.

Everything is working for smallest query.

Your help will be very appreciated to solve my problem.

Best regards,

Ben

Sherry George

Did you check the parameter values in NQSconfig.

MAX_ROWS_PER_CACHE_ENTRY

NQSConfig.INI File Configuration Settings - 11g Release 1 (11.1.1) . Also look at the best practices for cache size.

[Deleted User]

At the risk of stating the obvious:

BenFreez wrote:Here the situation, i have a report with more than 500 000 rows in the resultset and the query is taking 24 min to be executed.

That is a seriously flawed design and using caching is in any case just "throwing a blanket over the issue" rather than solving it.

Parsing 500k for an analysis? Sure.

Returning 500k for an analysis? Extremely poor choice.

There's even an official Oracle document detailing why this is a poor approach and what your alternatives are. Will update as soon as I have the link...it's lieing around here somewhere.

https://support.oracle.com/rs?type=doc&id=1558070.1

BenFreez

Hi Christian,

I kow this is not a best practice to hace that questity of data in resultset but it's a requirement from the business. I'am trying to find the best way to solve this problem

Benoit

[Deleted User]

Not it's not a requirement. It's a senseless implementation of a need.

Have you looked at the document?

If your requirment "i need 500k rows in a csv" then the solution is NOT "hey lets drag all this data from DB over network to the BI server to BI presentation server over the network to the browser and then from the browser to the desktop."

In terms of inefficiency you'll win any contest with that.

How about thinking about the basic need on a lower level and solve it in an actually efficient way?