What is the most efficient collection for accessing an object based on two key variants?

Question

I'm trying to implement a large cache of objects (up to 500000) and need to be able to access them in two different ways...

The key for each item is made up of three different strings; ItemNumber, PubCode and SizeCode. In some circumstances I'll be calling for a match on all three of those values to return a single object. In other cirumstances I'll be calling for a match on only ItemNumber and PubCode with a view to returning a collection of objects.

What is the best collection to use for this?

I've considered just using a generic list of object (of which all three of the key values are properties of) and using LINQ to query it, however I don't believe this is going to be the most performant way of doing this especially when you consider the size of the collection.

Any help will as always be appreciated!

Answer 1

How many items are you likely to have for any ItemNumber / PubCode combination? If the answer is "reasonably few" then I would start off with a Lookup<ItemNumberPubCode, Value> (or a Dictionary<ItemNumberPubCode, List<Value>>) - so if you're asked to look up by just the two of them, you can get straight to all matches. If you're asked to look up by all three, you fetch all the matches of the first two really quickly, and then do an O(n) scan for any match by SizeCode.

(Here ItemNumberPubCode is a type encapsulating the ItemNumber and PubCode; this could be an anonymous type, a Tuple<string, string>, or a real type.)

If there can be lots of matches for a particular ItemNumber / PubCode combination, then you might want a Dictionary<ItemNumberPubCode, Dictionary<string, Value>> - that will let you efficiently search by all three, and from just two of them you can fetch the dictionaries and use the Values property to find all matching values for the pair.