In distributed systems like SaaSGrid, storage of non-executable content that is used by applications during runtime must be centralized, and/or the functionality to access that content must be abstracted. In distributed systems, you must be node-agnostic, as you cannot guarantee the path a request will trace through the grid, nor what nodes are participating in the grid at any given time.

Some common options here include use of common network fileservers or SANs (\\server\share), SQL Server BLOB columns and cloud services like Amazon's S3 and Azure's BLOB Storage Services.

Below is a diagram representing the interactions between website application end users and IIS, IIS and WCF, and IIS, WCF and potential datastores.