As with “All Things Cloud” for the last year or so, Cloud BI is a subject of much buzz. But if you look around, aside from a few niche vendors that have developed Software as a Service (SaaS) and Platform as a Service (PaaS) BI packages, there’s not a lot of clear information or even thinking. Something close to home that still feels somewhat up in the air is implementing a BI solution in Windows Azure.
Yes, there’s technical information about Azure out there on the interwebs — mostly targeted at application developers and Hyper-V administrators. Even as full installations of SQL Server in the cloud have become possible via the Infrastructure as a Service (IaaS) feature of Azure called Azure VMs, there has not yet been a bloom of Cloud BI solutions being built in Azure or writing about said bloom. So I wanted to collect the information that’s out there and try to form a coherent picture of the current possibilities for BI in Azure.
To begin with, let’s clarify what we’re talking about with the term “Cloud BI,” and why it exists. A Cloud BI solution is basically a BI solution that has a data source, analytics processing, a delivery platform, or one or more of the above, housed in the cloud. If only one of the preceding components is in the cloud, it’s technically a “hybrid cloud” solution. An example of this in the BI realm might be a pre-packaged end-to-end Analytics platform for a particular vertical market. Or it might be a “Data as a Service” (DaaS) application providing customers with access to data of interest along with built-in analytical tools/reports. Moving BI into the cloud would be desirable for the same reasons as any cloud application: reduced infrastructure overhead and easy scalability. Since BI can be resource intensive — and therefore hardware intensive — the enticement of reducing server purchase and maintenance costs can be very attractive.
But having defined our terms, I return to my earlier question: what does Windows Azure offer as a platform in this arena? To begin with, it has the flexible pay-per-use cost structure that initially attracts decision makers to the cloud. Beyond that, Azure has several different data services offerings aimed at different use cases. On the PaaS (Platform as a Service) front, Azure offers a cloud-based version of SQL Server called Azure SQL Database (formerly known as “SQL Azure”), a “NoSQL” storage service called Azure Tables (primarily intended for non-relational data), and a Hadoop “Big Data” platform called HDInsight. On the IaaS (Infrastructure as a Service) front, hosted virtual machines are offered via Windows Azure VMs. Azure VMs can be built using any of the last several iterations of the Windows OS (up to and including Windows Server 2012), which of course supports SQL Server 2012 and SharePoint 2013, and can be configured to tie into existing Active Directory installations.
So Azure definitely has the theoretical facilities to run BI and other data services in the cloud. But the path to a BI solution must take into account some compromises and limitations inherent in the cloud platform, and in the Azure offerings themselves. Most of those limitations have to do with how much data you are working with, whether it’s structured or unstructured, and how you want to use it. One really needs an explanation of what use cases match each Windows Azure data services offering…
But that breakdown will have to wait until next time, when I’ll dive a little deeper into the details of Azure SQL Database and Azure Tables, and I’ll also talk a little bit more about HDInsight compares to the other Big Data offerings in the Microsoft ecosystem.
And in the next several posts, we’ll look in more depth at running full-on SQL Server BI on Azure VMs, at what limitations you might run into, and what user-facing features and capabilities go along with the “Microsoft Cloud BI story.”
