A guide to building databases in the cloud

Is a cloud based database right for me?
Businesses can deploy cloud hosted databases with a cloud services provider as a standalone product in a Database-as-a-service (DBaaS) model, but increasingly database services are being bought as part of a larger Infrastructure-as-a-Service solution.

CIOs who take an ‘everything in the cloud’ approach should think carefully before putting their database into the cloud. It is important that cloud isn’t seen as the fix-all solution to database management. One of the main limitations of putting databases in the cloud is that it isn’t yet suitable for hosting very large databases. A physical infrastructure is still the best for mass scale until cloud technology is able to match its capabilities. This is especially true of databases that support mission critical services.

CIOs who do want to take the cloud approach for their databases should think about deploying DBaaS services that are built ‘wide not tall’. This means rather than adding more data ‘vertically’ into one central database it would be better to grow the database ‘horizontally’ by linking multiple smaller databases built across a cloud infrastructure.

Build wide not tall
There are multiple benefits to this approach. Firstly using multiple smaller databases means that services are easier to keep running at optimal speeds, which could be a problem for larger, more cumbersome vertically built databases. This approach also enables businesses to cover multiple geographic locations, an advantage for many international enterprises.

The traditional approach to fixing a slow running database is to throw more hardware at it. However, this approach doesn’t work as well in the cloud. As with all cloud based services it’s not merely a case of adding more RAM to an under-performing application. Databases that have been built using the ‘wide not tall’ principle are far easier to scale up. When a database has reached capacity another server can be spun up to create a new database. Because the application is cloud-based this process is often as simple as logging onto an online portal and spinning up another server in a few clicks.

Spinning up servers as and when needed also allows businesses to have greater control over IT costs. The ability to scale a database up and down means that there is no need to keep running servers, and paying for them when they are not being used – one of the most commonly cited benefits of the cloud.

Relational or NoSQL?
One of the big industry questions in relation to cloud databases is whether to build them using a conventional relational model or NoSQL. Looking at the issue from a cloud provider’s point of view, it isn’t necessarily easy for them to deploy Oracle for cloud databases. The reason it doesn’t work particularly well with the cloud because its licensing requires dedicated hardware. This isn’t compatible with cloud systems built on shared multi-tenant platforms, which is the model for many providers, including NaviSite.

The new generation of databases, which are built on the NoSQL model, currently represent a small percentage of the market but is growing. They are designed to suit the horizontal scale model and deal with multiple small requests. In this way they are much better suited to cloud applications.
Companies like Google and Facebook are back-ended with this new generation of databases. So for those who have written applications specifically for cloud, they will use one of these types of databases. This is one of the reasons the NoSQL market will continue to grow. Also as new programmers leave college and university they will start writing for these new platforms. Oracle, Microsoft SQL etc. are recognising a shift and are doing their best to adapt and they are beginning to do some interesting things which means that in 10 years both types of database are likely to still be around.
As CIOs look to the cloud to enhance their business’s flexibility and scalability, cloud hosted databases will increasingly feature in the suite of IT services offered by service providers. Built on a horizontal model these databases – both SQL and NoSQL – will continue to evolve and become more appealing to businesses looking to migrate existing databases or build in the cloud.