Best implementation for large datasets

[C# Programming]

I am currently building an application which will deal with several large sets of data from different sources such as databases, csv & xls files. I have been looking at the dis/advantages of using dataset & generic collections.

My conclusion was that I would need to use generic collections because I will be dealing with ~ 300,000 records in most cases (they will not usually be displayed to the user). I have not really delved into generic collections before and I am having trouble figuring the best way to construct the class because each set of data will have a different amount fields but I want to be able to use the same methods to sort, search & filter.

Would the below basis for a class work, is there a better way? I am having trouble thinking how the sorting would work with this, whilst it seems very easy when the class is not made up with an array.

public class Record
{
private string[] cells;
public string[] Cells
{
get { return cells; }
set { cells = value; }
}