Wednesday, August 25, 2010
I'm reading: How to select the best cloud environment for your business intelligence solution ?Tweet this !
Amazon EC2, Rackspace, Salesforce, GoGrid are some of the famous cloud providers, and Microsoft Azure is probably the newest kid on the block. Cloud is gaining popularity day by day, and businesses as well as solution providers would want to move their existing applications or base their future applications on the cloud. But there are a few factors which should be considered for evaluating different cloud providers that suit your needs. Below is a brief list of such requirements in alignment with BI / MS BI needs.
1) Platform migration without architectural changes : Applications that are already developed when needs to be migrated to cloud, change in architecture of the application due to limitations or constraints of the cloud vendor is out of question. For this requirement the cloud vendor should be providing IaaS and not just PaaS services. The reason I mentioned PaaS and not SaaS is that if one is considering to use SaaS , this would make sense for your future requirements but not for your existing applications. Amazon EC2 provides a major set of services to cater this requirement. Unfortunately, to the best of my knowledge, Microsoft Azure is not up to the mark till date for this requirement.
2) Support for Private Cloud : There might be some very sensitive business logic in WCF (Windows Communication Foundation) services or similar other interfaces that one might not want to expose on the cloud. So the cloud vendor should also be supporting private cloud like Amazon VPC or Windows Azure Appliance Solution.
3) Software Licensing : Many cloud providers do not facilitate use of corporate licenses that enterprises would have procured, and many software vendors do not provide license for the use of product on cloud. Licensing needs to be levelled from software vendor as well as cloud vendor, so that software licenses can be easily used / reused on cloud environment.
4) PaaS Support for operations related to data : In a typical BI project, involving ETL, one can expect more than one database and different forms of staging ( in different file formats ). Cloud vendor should be supporting multiple relational DBs and storage formats required to support data storage. For example, SQL Azure and Project Houston provide a nice platform for data storage, design and operations. Windows storage provides three different kinds of storage formats. But in case if both are required to operate in the same environment, where I need to pull of data from SQL Azure and store in a file format provided by Windows Azure ( say BLOB storage ), both are not on the same platform. If your project has such dependencies, this should be taken care before considering your cloud vendor.
5) Ease of backup and restore operations : Most cloud vendor provides features that flush out any data as soon as you stop paying for the instance i.e using the instance. For permanent data, a separate dedicated storage needs to be purchased. Backing up and restoring this instance can be one big concern. Also enterprises might want portability for such instances, where one might want to create images on the cloud and use it for multiple projects or solutions. Amazon Machine Images is one such example in alignment to this requirement.
Please note that I have considered Amazon for the sake of discussion, but by no means this implies that Amazon EC2 is the best cloud vendor for your requirements. It always depends on the requirements of an enterprise, billing rates and services provided by cloud vendor in accordance with the requirements of the enterprise. The above listed factors should be evaluated at early stages before adopting the cloud environment and considering a cloud vendor.