LINQ-To-SQL - 慢查询

Question

I'm just wondering if anyone can offer any advice on how to improve my query.

Basically, it'll be merging 2 rows into 1. The only thing the rows will differ by is a 'Type' char column ('S' or 'C') and the Value. What I want to do is select one row, with the 'S' value and the 'C' value, and calculate the difference (SC).

My query works, but it's pretty slow - it takes around 8 seconds to get the results, which is not ideal for my application. I wish I could change the database structure but I can't sadly!

Here is my query:

var sales = (from cm in dc.ConsignmentMarginBreakdowns
            join sl in dc.SageAccounts on new { LegacyID = cm.Customer, Customer = true } equals new { LegacyID = sl.LegacyID, Customer = sl.Customer }
            join ss in dc.SageAccounts on sl.ParentAccount equals ss.ID
            join vt in dc.VehicleTypes on cm.ConsignmentTripBreakdown.VehicleType.Trim() equals vt.ID.ToString() into vtg
            where cm.ConsignmentTripBreakdown.DeliveryDate >= dates.FromDate && cm.ConsignmentTripBreakdown.DeliveryDate <= dates.ToDate
            where (customer == null || ss.SageID == customer)
            where cm.BreakdownType == 'S'
            orderby cm.Depot, cm.TripNumber
            select new
            {
                NTConsignment = cm.NTConsignment,
                Trip = cm.ConsignmentTripBreakdown,
                LegacyID = cm.LegacyID,
                Costs = dc.ConsignmentMarginBreakdowns.Where(a => a.BreakdownType == 'C' && a.NTConsignment == cm.NTConsignment && a.LegacyID == cm.LegacyID && a.TripDate == cm.TripDate && a.Depot == cm.Depot && a.TripNumber == cm.TripNumber).Single().Value,
                Sales = cm.Value ?? 0.00m,
                Customer = cm.Customer,
                SageID = ss.SageID,
                CustomerName = ss.ShortName,
                FullCustomerName = ss.Name,
                Vehicle = cm.ConsignmentTripBreakdown.Vehicle ?? "None",
                VehicleType = vtg.FirstOrDefault().VehicleTypeDescription ?? "Subcontractor"
            });

Answer 1

There really isn't enough information supplied to make an informed opinion. For example, how many rows in each of the tables? What does the generated T-SQL look like?

One thing I would suggest first is to take the outputted T-SQL, generate a query plan and look for table or index scans.

Answer 2

A good place to start when optimizing Linq to SQL queries is the SQL Server Profiler. There you can find what SQL code is being generated by Linq to SQL. From there, you can toy around with the linq query to see if you can get it to write a better query. If that doesn't work, you can always write a stored procedure by hand, and then call it from Linq to SQL.