JavaScript Object Notation (JSON) is the data structure of the Web. It's a simple data format that allows programmers to store and communicate sets of values, lists, and key-value mappings across systems. JSON is really just a simple schema less language that was created to support web pages. If your data is small and simple then a JSON document database may work OK. If you have more connected and large data sets, it is not a good solution. When data gets more complex, larger and more connected, an object or graph database will get the job done much better with way better performance and scalability. First, JSON is very limited in data types supported, it only supports 4 data types: Number, String, Boolean and Array. Secondly, it is usually schema less meaning that you can store any JSON object without following any rules of what fields or field types each object contain. That means that each JSON object enters a database as a text string that is parsed to discover its fields and types. JSON does not support references between objects but references are often accomplished by designating one of the string fields to hold an object id using UUID text strings such as "6438850b-b33d-475f-95d7-8997606431a5".

Now compare this to how it is done with VelocityDB; Unlimited data types supported, almost any .NET data type can be persisted (one exception is Delegates as it involves code pointer data). VelocityDB uses a schema so that we can control what is stored. Any data, even unstructured, can still be stored as you can control how specific you make the schema. Having a schema enables serializing .NET object without having to describe each field data type, field name for each object stored. Redundancy is eliminated and objects can be used and stored as binary data with just the field values, not all its meta data, for each object. Having the schema gives structure that is used when constructing Linq queries. Support for object references is built in and only occupies 4 - 12 bytes in persistent storage. You can still use JSON when interacting with Web applications, Microsoft supports .NET object to/from JSON when used with frameworks such as Entity Framework. VelocityDB support import and export to/from JSON. Anything that can be done with a dedicated JSON database can also be done with VelocityDB. Simple comparisons between VelocityDB and JSON databases shows how performance and data size is affected by these differences. As data get larger these differences will increase. See comparison with MongoDB and Couchbase.

A relational database stores all data in tables on a server. Clients talk to such servers using SQL statements such as "select * from actors". Data is also created and updated using SQL. VelocityDB is not a relational database (but it is very good at managing relations!), with VelocityDB C# objects are stored as objects with all references/relations between the objects. Using VelocityDB persistent C# objects is very similar to using in memory C# objects. Data is created by creating C# objects; updates are accomplished by updating C# objects. The persistent store is safely done using transactions with locking protection so that one user can't accidently undo other user's changes. Like most database systems, VelocityDB uses paging but unlike other systems VelocityDB data pages have variable size and can optionally be encrypted and compressed. VelocityDB persistent objects have an object identifier (Oid) consisting a DatabaseNumber-PageNumber-SlotNumber.

A detailed comparison between VelocityDB and a relational database can be found here as a pdf doc, as html here or original Word document here. This is a document under construction. Help us if you can.

The most important benefit is the flexibility that this database system provides. No other types of database system lets you fine tune your application to the finest details. Using VelocityDB, you can choose the best possible data structures for your application. You can control where you place the data persistently and how it's indexed and accessed. With VelocityDB you will have the ultimate tool for persisting data efficiently with unbeatable performance.

Unstructured Data can be managed efficiently using VelocityDB. Several of the sample applications provided with the download explore unstructured data. The triangle counter test demonstrates how efficient it is to follow association/relations. In memory, an association in VelocityDB is simply a C# object reference. It can't get any more efficient than that. The Kevin Bacon Numbers sample is another example showing how efficient we can handle unstructured data. A common way to manage unstructured data is to add such data to an inverted index. That way we can efficiently query such data, like if we want to know what store products that match one or more words. The entire Google search mechanism is based on an inverted index. In some of our samples we use inverted indexes, these samples include; TextIndexer and Wikipedia.

A shopping web site in prototype production uses an inverted index to help shoppers find products they are looking for, try the product search here (it is not always online).  It currently manages about 30 million products from about 650 stores (~33 GB of data).

VelocityDB is extremely scalable; it can handle small applications with small amounts of persistent data as well as applications using trillions of objects up to about half a Yottabyte in size per session. Unlimited number of sessions can be used per application. Performance is fantastic, a billion Person object database is quickly created, less than 1 hour, and can be browsed with the Database Manager.