In my last post in this series, I described the basics of Data Services within Microsoft’s Windows Azure public cloud platform. Basically, we were able to somewhat exclude the majority of PaaS offerings in Azure and home in on the IaaS offering: Azure Virtual Machines.
Azure VM’s are effectively the only real path to full-fledged Microsoft BI in the cloud. This is because VMs provide the only way to leverage the BI tools included with SQL Server. Azure SQL Database is primarily the Database engine as a service, and does not offer any form of Integration Services or Analysis Services as a platform service. So if you want any kind of SSIS or SSAS features, you have to be able to run full-blooded SQL Server. Really, you can also include Azure SQL Reporting Services in the same group, as it can only report against an Azure SQL Database source – so it’s use in the BI solution might be limited.
Another compelling reason to look at Azure VMs has to do with SharePoint. SharePoint is, in some incarnation, the endorsed front-end/web delivery platform for Microsoft BI. So it’s a fair bet that if you are running Microsoft BI you are also running — or considering running — SharePoint. For parties looking to migrate an existing SharePoint BI solution to the cloud from on-premise, you can install SharePoint on Azure VMs, and run them virtually networked with your BI virtual machines. So your entire SharePoint BI solution can be hosted in the cloud on Azure VMs. In the near future, Power BI promises to revamp that story as Office 365 and SharePoint Online fully penetrate the market. So, for new installations it’s probably wise at this point to look into Microsoft’s PaaS offerings for cloud-based SharePoint.
So when we say this stuff about VMs, what exactly does it entail? At the highest level, you virtualize your BI solution servers, and host them in Azure. You can either upload existing VMs, or create them via the Azure Portal interface. You can keep the VMs in your AD farm even though they are deployed in the cloud, and you can virtually network them together to avoid round-tripping back to on-premise resources. But, basically, you have a virtualized solution. What do you gain in this move? First, you gain back all of your BI/SharePoint solution hardware and related costs. Second, you can implement horizontal scaling to allow the system to automatically throttle bandwidth and storage capacity as needed (note that this is not automatic behavior…). And third, you can increase productivity by providing development personnel access to the system from any browser, anywhere.
So with Azure VMs you can run full-on SQL Server BI while reducing costs, improving scaling, and increasing productivity. What’s not to like? Well, there are limitations and gotchas (of course). This is the real world, right? Next time, we’ll discuss more specifics of Azure VMs and specific concerns about running BI on them.
Microsoft Cloud BI: Azure Virtual Machines, pt. 1
by Andrew Tegethoff on August 20th, 2013 | ~ minute read
