How to reduce many similar correlated subqueries?

问题

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

回答1:

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  

回答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.

来源：https://stackoverflow.com/questions/7592692/how-to-reduce-many-similar-correlated-subqueries