BinaryRage - the ultra fast .NET key/value store

• BinaryRage is designed to be a lightweight ultra fast key/value store for .NET with no dependencies
• It's production-ready - already in several large production-environments
• Supports complex objects out of the box (and lists of objects)
• 100% FREE and open (no paid pro version or anything lame like that)
• No configuration, no strange driver/connector, no server, no setup - simply reference the dll and start using it in less than a minute.
• Created it because I think there is a huge need for a very simple key/value store
• If you hate writing boilerplate code - you will love BinaryRage

Created by Michael Christensen (shameless facts)

• I'm Chief Technical Officer, Technical Lead in a mid-sized web-agency in Denmark using .NET
• Doing a bit of consulting on special occasions
• Love startups and performance
• Twitter: @mchidk

Show me the code

Simple class - simply include an [Serializable] attribute.

[Serializable]
public class Product
{
	public string Title { get; set; }
	public string ThumbUrl { get; set; }
	public string Description { get; set; }
	public float Price { get; set; }
}

Insert-syntax (same for create and update)

BinaryRage.DB<Product>.Insert("mykey", myProduct, @"C:\testpath");

... or with a list

BinaryRage.DB<List<Product>>.Insert("mykey", listOfProducts, @"C:\testpath");

Get the saved data

var myProduct = BinaryRage.DB<Product>.Get("mykey", @"C:\testpath");

... or with a list

var listOfProducts = BinaryRage.DB<List<Product>>.Get("mykey", @"C:\testpath");

Query objects directly with LINQ

var bestsellers = BinaryRage.DB<List<Category>>.Get("bestsellers", @"C:\products\").Where(p => !string.IsNullOrEmpty(p.Name));

That's it - can it be any simpler?

Key API

Generate a unique key

BinaryRage.Key.GenerateUniqueKey()

Calculate checksum on a string

BinaryRage.Key.CalculateChecksum(string inputString)

Fast MD5Hash generating

BinaryRage.Key.GenerateMD5Hash(string input)

Goals

Codebase and usage must be as simple as possible (but not simpler).

Todo

• Always check out open issues if you want to help out
• Handle very large objects

FAQ

Is it really fast?

We have tested more than 200,000 complex objects written to disk per second on a crappy laptop :-)

All writes are performed asynchronously. Reads are instantly available (also if writes are not completed) - by design.

Why are you compressing the objects before written to disk?

The less I/O - the better. A compressed object is doing less I/O because of fewer bytes written to disk - simple as that.

Why do I have to provide a path when I save an object?

I have decided that I want to provide sharding directly from the start. You can easily wrap the insert and get methods and "hardcode" the location if you want.