There is a shift going it the IT world: more and more applications are moving to the cloud. Recent examples, e.g. Office365, shows that the gap between the on-premise version and the cloud version are rapidly closing. Even the promise of Microsoft to put more and more effort in the development of cloud application: every Microsoft product nowadays is ‘cloud-aware’. The big question is: what about BI and the cloud. One thing I know for sure: it will come. Not completely sure is how, but I have an idea and that is the main reason why I started this blog and write this post.
The recent announcement of Microsoft the upcoming beta of SQL Azure Reporting shows the availability of Cloud BI. Also the planned release of BizTalk in the cloud hints to Integration Service in the cloud. The only main part that is missing in the cloud from the on-premise SQL Server is the Analysis Service part. But I expect that even that will move to the cloud. With the announcement of Crescent in the upcoming release of SQL Server (Denali) a big step is made.
So there are multiple hints/announcements that hints to a Cloud BI solution platform. But with a platform you don’t have a working solution. So how do we, as BI consultants, make a) those solution and b) make them work for our customers? The first part is not that difficult if we have the tools and part of our job. The latter is more complex because we should break with our BI traditions.
In my option the idea of a full ETL process with Staging and an EDW will not be doable in a cloud solution. What is the purpose of transporting the data over and over again? As more and more application will life in the cloud, so is the data. And basically the main idea of cloud storage is that it doesn’t matter where it is physically stored. That is why we don’t need a complete ETL process in the cloud. Probably a more simple solution. like the SQL Azure Data Sync, but then autonomic and (near) realtime. The end point will be data-adapter which we can define a virtual data warehouse. All these data-adapters we can publish in a data-market, like the Windows Azure Marketplace DataMarket. There will be a mix of pubic data and private data, which will enhance the value of the data.
As the virtual data warehouse is not a real data warehouse, but a definition of how and which datasets are linked together the need for an analytical platform is still there. With that analytical platform, I thinking of a cloud interpretation of PowerPivot, we can feed the reporting environments, like a Report Builder or the newly data visualizing tools codenamed Crescent. All this not yet available and the main invent is lying in the integration part and mainly on how to create datasets and link them together, but until then I keep my eyes open.