Question

I have a scenario where I get a count and then pass the count as a variable to a similar query to get the paginated records. So basically I am doing a full query to get all the count by internally creating the full table and then using that count to display the same table with 10 per page. What solutions do I have to avoid this sort of multiple query?

Something like this is a Pseudo language .

select count {big table}

select big table where records are between count and count+10

Is there a sensible way to get the COUNT variable in the same query?

I am wondering how would Google handle a search, would it first find all the records or just fetch the records without tracking the no: of pages? Page numbers can’t be computed prior as it is dependent on the variable sent by the user.

Edit: I have a similar question here https://dba.stackexchange.com/questions/161586/including-count-of-a-result-in-the-main-query

Answer 1

Regarding Google, they are likely to generate only the requested amount of results (like 10) and to estimate the count. The estimated count is very imprecise.

You can’t have SQL Server count all results and get only a subset of them. There 3 strategies to deal with this:

• execute a counting and a data query
• execute an unlimited data query and discard all but ten results on the client
• execute an unlimited data query into a temp-table whose primary key is the row number. You can then count instantly (get the last row) and select any subset by rownumber with a single seek

Counting the data can be significantly cheaper because SQL Server can use different indexes or discard joins.