A first glance at SQL Server Data Services

I’ve been researching one of Microsoft’s newest services: SQL Server Data Services (aka SSDS). In this post I will present my first impressions about this technology.

Motivation

The motivation of this service is to provide a scalable data storage and query processor utility "on the cloud". Using this approach, the costs of infrastructure and support decreases, since that responsibility relies on Microsoft and not in the software provider.

Besides, more and more new Web applications need to save less complex data, with a high service level agreement (SLA).

Implementation

SSDS is built on the SQL Server technology, using the SOAP and REST communication protocols. Some of the features are:

• No schema. Since the data model is flexible, we now can have heterogeneous entities (serialization becomes more complex).
• Pay-as-you-grow model. The space used can grow as you need it.
• SLA covering high availability, performance and redundancy.
• Easy to develop. In order to use the data we only need to call the appropriate Web Service.
• The queries are text-based with a C# LINQ pattern.

The organization model is really simple:

• Customers are the organizations that are using the SSDS. Customers have many Accounts.
• Accounts are the billing entities. An account has authorities assigned.
• Authorities are the ones that organize the data containers. It defines the location and security policies of its containers.
• Containers are the scope for the entity storage and query. For example, each employee may have a container with its personal data.
• Entities are the storage unit of SSDS, for example the employee’s education information.

Conclusion

I think that this service has a lot of potential. With the growth of Web2.0 and social applications, we need a way of deploying small applications with a reliable data access. Obviously is very expensive to deploy and maintain a data base server, so having that service on-the-cloud is a very interesting alternative.

From the technical point of view, the no-schema structure is very interesting. Still, I guess that it can have a negative impact in the serialization process.

On the other hand, this service is still in working process, so we’ll be awaiting for new features to come.