Thanks for the quick reply. Well, what about Oracle Ultra Search. As far as I am concerned it is a solution for indexing and searching the enterprise documents and web site, and it is for the enterprise internal use only. Is this right?

Is it (Oracle Ultra Search) is the same as </code> http://www.google.com/enterprise <code>

Regards,
Emad

Hi Tom,
Any idea what is BIN$RYy1xZLSQqKtyjg2mBJfUQ==$0 trigger on wk_test.

I intalled my 10gr2 database manually and I got this trigger, so I recreated the database again, to be sure there was no mistake in installation.
And beacuse this problem persisted I decided to drop, meanwhile I'm not using ultra search options.
And everything is recompiled.

But do you have some idea what does it means?
Thanks

Thanks

Hi,

So from my understanding, there is no tool to query a database directly from a google search.
I mean for example:
1- type in google: find an appartment in london
2- parse this request in a website that would contain an open database to the web.
3- send back the answer

Am I right ?
Thanks

Hello Tom,

I have recently reading about how to have Google friendly pages. However, I have read the following link

http://blogs.oracle.com/otn/2007/02/08#a375

Is not Oracle Docs were google friendly?? If not, what modifications have been done to make it so.

This is crucial to us since we are posting at our websites many PDFs and DOCs and we need Google be able to index it in addition to the normal static and dynamic HTML pages.

One more thing please, we seriously thinking to have CMS (Content Management System) to store our rich files and manage it. I am kindly asking to point me where I have to start here since we need one that is integrated with Oracle 10g.

Thanks in advance

Hi

I have an application supporting citizens in a city (quite big, around 2 million persons).
There is a query in the main search engine which usually returns more than 20000 rows. No matter what I try to speed up the query (it takes around 3 seconds to run, quite long for a search engine) I cannot getter better than that.

The query's predicate contains 6 conditions and I have a composite index for those 6 conditions (6 columns), the most restrictive query returns 20000 rows as I mentioned, it is tiny in a 2 million rows table but index scan of 20000 rows isnt very good.

I wonder is there some clever way to get this faster? I know it's hard question but I have been trying to getter better for the last 2 months without much success. I have employed keep pool, increase buffer cache, get faster CPUs.

Thanks a lot

Hi

do you use all six conditions every time?

-- most of the times, sometimes 4 or 5 but not often

is the table "read mostly" (eg: relatively static) or very dynamic?

-- it's has inserts and updates but few, the load are batch processes running during nights

do you need to get 20,000 records or is the real goal "get the first 25, show them and tell them 'there be MORE records if you hit next!!!". I don't believe ANYONE looks at result 1234 - ever.

-- I understand what you mean however even I use query such as

select *
from (select rownum ron, x.*
from (select * from t1
where ..... 6 conditions
order by y) x
where ron <= 1000) z
where z.ron >= 925 -- 25 per page

the inner query X still needs to walk the index 20000 timees correct? Which means same number of consistent gets no?

Functionally there are some citizens who likes to look backwards, instead of looking forst page they actually click the last and start from there but I guess we can ignore this sort of behaviour and returning only 10% of actual rows.

Thanks Tom for this quick reply.

I meant by Google friendly the following

http://www.google.com/support/webmasters/bin/answer.py?answer=40349

Quickly reviewing Oracle Content Database, it sounds great.

As usual, can not find words to thank you.

Hi

I dont get the comment

"the inner query need not "walk" the index 20,000 times - no. It depends."

under what circumstances inner query doesnt have walk the index 20000 times? I mean if i want to get an ordered result set I must get the whole result set (therefore 20000 index scans) and order it correct?