Comments by glitteringsound (Top 4 by date)

Previously I used entity framework for saving 100K rocords in database where each recod further contains child records.

my object hierarchy is :

public class Consignment

{

....

List<consignmentaddress> addresses {get; set;}

List<consignmentline> lines { get; set; }

}

I could expect Consignments Count = 100K and each Consignment could contain 2 ConsignmentLines and on average 3 Consignment Addresses.

So this all makes roughly = 100K + 200K + 300K = 600K records .

For inserting all these 600K records, I used entity framework and this was damn slow. Then I devided the file in chunks and parse 1000 records and inserted using EF by recreating context each time. This was very improved . But still that does not full fill my requirement ( as in future i could expect 500K root level records and each further could contain 3-4 child record in each list. That could make 1200K records in database).

Now I need fast and efficient solution.

That's why I want to send the whole hierarchy at once, so that i can iterate that collection inside SQL and insert the parent child records. As this would be close to metal , it would be 10 times faster than entity framework.
But at the moment, I am iterating this List<consignment> collection from .NET C# in loop and for each consignment, along with it's children ( i.e List<consignmentaddress>, List<consignmentline> ), I am calling stored procedure and sending these 3 as TVP ( 1 for Consignment, 2nd for List<consignmentaddress>, 3rd for List<consignmentline> ). This procedure is also very slow. Because, for 100K root records, 100K stored procedure calls are going.

Now I need a solution, which takes the whole hierarchy or at least some chunk and push all the data at once by keeping whole parent child relationship intact.

Ok I googled. This will create directories on my disk here. What it has to do with my problem? How creating the directory will actually solve my problem?

What do you mean by directory class? How can it be implemented?