Graph DB's Vs Azure Table storage for a Social networking application

I'm starting on some architecture work for a .Net based social networking application to be hosted on Azure cloud. we are going to be using ASP.NET MVC on the front end.

i would like to consider the options for storage. considering scalability needs and due to the inter-connected nature of the application, SQL azure has been ruled out.

what would be the main considerations in choosing a graph DB such as Sones GraphDB or neo4j which have features s开发者_Python百科pecific for a social networking application against using windows azure table storage to achieve the needs.

i'm mostly concerned about development time, cost, ability to leverage existing skills like .NET and reliability of the graph DB platforms and ease of setup and administration.

Graph databases are designed for applications such as social networks. For ease of development, it may be best to start with something like GraphDB. A key advantage over a key-value database is powerful query and traversal capabilities. It would be easy to, for instance, find all occurrences of friends of friends using the GraphDB query syntax.

The benefit of a key-value database service like Azure Table is low cost, minimal administrative overhead and scalability. You can store 500TB of data per Azure storage account and setup accounts in multiple regions. There is no server setup or database administration overhead and the Visual Studio SDK is easy to use. The down side is that graph like query support is not built in and you must index off the Primary Key / Row Key pair. For additional Azure Table design pattern please see https://azure.microsoft.com/en-us/documentation/articles/storage-table-design-guide/