How to reduce many similar correlated subqueries?

Question

This is part of a larger statement, but I'm wondering if CTE or another method would help make this more efficient or cleaner. I could write it as a table-valued function and include it in my from clause, but I'd like to avoid extra objects if there is another solution.

The SELECT TOP 1 ... sub-queries here simply catch when I have a rate with an earlier effective date than the base table, but I'm not fond of repeating them for each column I need to access. Is there a better way to accomplish this, or is this a normal looking statement?

SELECT j.EmployeeId
       ,j.CompanyId
       ,j.JobCode
       ,COALESCE(j.CustomWageRate, r.WageRate, (SELECT TOP 1 WageRate
                                                FROM   ContractLabor.CompanyJobRates
                                                WHERE  CompanyId = j.CompanyId
                                                       AND JobCode = j.JobCode
                                                       AND EffectiveDate < j.EffectiveDate
                                                ORDER  BY EffectiveDate DESC), 0) AS EffectiveRate
       ,COALESCE(r.CustomBurdenRateReg, (SELECT TOP 1 CustomBurdenRateReg
                                         FROM   ContractLabor.CompanyJobRates
                                         WHERE  CompanyId = j.CompanyId
                                                AND JobCode = j.JobCode
                                                AND EffectiveDate < j.EffectiveDate
                                         ORDER  BY EffectiveDate DESC)) AS CustomBurdenRateReg
       ,COALESCE(r.CustomBurdenRateOvt, (SELECT TOP 1 CustomBurdenRateOvt
                                         FROM   ContractLabor.CompanyJobRates
                                         WHERE  CompanyId = j.CompanyId
                                                AND JobCode = j.JobCode
                                                AND EffectiveDate < j.EffectiveDate
                                         ORDER  BY EffectiveDate DESC)) AS CustomBurdenRateOvt
       ,COALESCE(r.CustomBurdenRateDbl, (SELECT TOP 1 CustomBurdenRateDbl
                                         FROM   ContractLabor.CompanyJobRates
                                         WHERE  CompanyId = j.CompanyId
                                                AND JobCode = j.JobCode
                                                AND EffectiveDate < j.EffectiveDate
                                         ORDER  BY EffectiveDate DESC)) AS CustomBurdenRateDbl
       ,j.EffectiveDate
FROM   ContractLabor.EmployeeJobDetails j
       LEFT JOIN ContractLabor.CompanyJobRates r
         ON j.CompanyId = r.CompanyId
            AND j.JobCode = r.JobCode
            AND j.EffectiveDate = r.EffectiveDate

Bogdan Sahlean · Accepted Answer · 2011-09-29T06:33:25.620

SELECT j.EmployeeId
       ,j.CompanyId
       ,j.JobCode
       ,COALESCE(j.CustomWageRate, r.WageRate, ca.WageRate, 0) AS EffectiveRate
       ,COALESCE(r.CustomBurdenRateReg, ca.CustomBurdenRateReg) AS CustomBurdenRateReg
       ,COALESCE(r.CustomBurdenRateOvt, ca.CustomBurdenRateOvt) AS CustomBurdenRateOvt
       ,COALESCE(r.CustomBurdenRateDbl, ca.CustomBurdenRateDbl) AS CustomBurdenRateDbl
       ,j.EffectiveDate
FROM   ContractLabor.EmployeeJobDetails j
       LEFT JOIN ContractLabor.CompanyJobRates r
         ON j.CompanyId = r.CompanyId
            AND j.JobCode = r.JobCode
            AND j.EffectiveDate = r.EffectiveDate

       OUTER APPLY --or CROSS APPLY
       (
            SELECT TOP 1 WageRate
                    ,CustomBurdenRateReg
                    ,CustomBurdenRateOvt
                    ,CustomBurdenRateDbl
            FROM   ContractLabor.CompanyJobRates
            WHERE  CompanyId = j.CompanyId
                    AND JobCode = j.JobCode
                    AND EffectiveDate < j.EffectiveDate
            ORDER  BY EffectiveDate DESC       
       ) ca

Ah, much better. `OUTER APPLY` was the missing piece I didn't know to look for. Thank you. — JustinStolle, Sep 29 '11 at 05:36
You're welcome. You could use `SET STATISTICS IO ON` to see how many logical and physical reads are done by this two queries. Less is better. [Using APPLY](http://msdn.microsoft.com/en-us/library/ms175156%28v=SQL.90%29.aspx). — Bogdan Sahlean, Sep 29 '11 at 05:52

Michael Petito · Answer 2 · 2011-09-29T05:34:31.817

2

You could join on a derived table expression in your outer query using cross apply / top 1 and select the relevant columns all at once.

Your query could then look like:

SELECT ..., ISNULL(x, defaultValues.x)
FROM ...
  CROSS APPLY (SELECT TOP(1) x, y, z FROM ... WHERE ... ORDER BY ...) defaultValues

It will be more efficient because there are fewer joins in the query (each of your original correlated subqueries turns into an outer join).

You can also use OUTER APPLY for outer join like functionality.

edited Sep 29 '11 at 05:34

answered Sep 29 '11 at 05:28

Michael Petito

12,891
4
40
54

Thank you. Yes, the `OUTER APPLY` works for me in this instance. – JustinStolle Sep 29 '11 at 05:44

How to reduce many similar correlated subqueries?

2 Answers2

Linked