Simple Order of Magnitude counts.. do they exist ?

Question

I don't normally deal with SQL, so apologies if what I am asking is dumb / obvious.
I have a table [orders] with approx 300,000,000 rows. It is quite wide (approx 500b on avg).
there is a unique index [idx_a] on [order_id , order_dt, store_chain, store_id]
there is another index [idx_b] on [order_dt, ...several other columns...]
At a high level I want to do something like:
SELECT TO_CHAR(order_dt, 'YYYY'), COUNT(*)
FROM orders
GROUP BY order_dt
But I am not really interested in the exact count, I am more interested in an order of magnitude summary. The system is online and live and continually accepting orders from the various outlets (approx 20 per minute). I have tried running the above query, in a test environment but it does not complete inside of 15 minutes.
Is there a smarter way to do this ?

Answer 1

Howler_Fish wrote:
At a high level I want to do something like:
SELECT TO_CHAR(order_dt, 'YYYY'), COUNT(*)
FROM orders
GROUP BY order_dt
But I am not really interested in the exact count, I am more interested in an order of magnitude summary. The system is online and live and continually accepting orders from the various outlets (approx 20 per minute). I have tried running the above query, in a test environment but it does not complete inside of 15 minutes.
Is there a smarter way to do this ?Apart from the other advices you've already got, you could try to use "sampling" if you're actually interesting in just a "rough" estimate:
SELECT TO_CHAR(order_dt, 'YYYY'), COUNT(*) * 10 as cnt
FROM orders sample (10)
GROUP BY order_dt;
This would "sample" only 10% of the rows using the "SAMPLE" clause, therefore you would have to multiply the count by the same factor to get results comparable to the "actual" query.
Please note that depending on how "skewed" the distribution of your data is across the table blocks if you use only a low sample size your results might be way off, so you might need to test with a different degree of sampling.
Regards,
Randolf
Oracle related stuff blog:
http://oracle-randolf.blogspot.com/
SQLTools++ for Oracle (Open source Oracle GUI for Windows):
http://www.sqltools-plusplus.org:7676/
http://sourceforge.net/projects/sqlt-pp/