LINQ-To-SQL : Efficient way to implement RANK()

Question

I'm trying to port the following piece of SQL to LINQ and I'm able to achieve what I need. But I believe there should be a better way to do it as the SQL generated as a CROSS JOIN, which is not performant.

SQL to port:

    SELECT bu_code,
       cust_no,
       cust_name,       
       RANK () OVER (PARTITION BY bu_code, cust_no ORDER BY bank_id DESC)    rnk
  FROM customer;

LINQ to port the above SQL:

from cust in customer
    group cust by new { cust.BuCode, cust.CustomerNumber } into grp
    let MaxBankId = grp.Max(g => g.BankId)
    from cust in grp
    where cust.BankId == MaxBankId
    select new
    {
      cust.BuCode,
      cust.CustomerNumber,
      cust.CustomerName                                         
    };

SQL generated by above query:

SELECT t3.BU_CODE, t3.CUST_NO, t3.CUST_NAME
  FROM (  SELECT t2.BU_CODE           AS "BuCode",
                 t2.CUST_NO           AS "CustomerNumber",
                 MAX (t2.BANK_ID)     AS C1
            FROM CUSTOMER t2
           WHERE     ( :p0 <> 0)
                 AND (NOT (t2.BU_CODE IS NULL))
                 AND (t2.BU_CODE = :p1)
        GROUP BY t2.BU_CODE, t2.CUST_NO) t1
       CROSS JOIN CUSTOMER t3
 WHERE     (t3.BANK_ID = t1.C1)
       AND (t1."BuCode" = t3.BU_CODE)
       AND (t1."CustomerNumber" = t3.CUST_NO)
       AND ( :p0 <> 0)
       AND (NOT (t3.BU_CODE IS NULL))
       AND (t3.BU_CODE = :p1)

PS: I'm working with Oracle 11g and Devart.

Thoughts on this?

Answer 1

Not all SQL constructions can be ported to LINQ. Not all LINQ queries are translated to SQL as efficiently as possible. In some situations, it is better to write a query in SQL and only materialize its result using an ORM. You're using LinqConnect, aren't you? Then you can use DataContext.Query (string query) method: https://www.devart.com/linqconnect/docs/?Devart.Data.Linq~Devart.Data.Linq.DataContext~Query.html .