Skip to Content

Cloud Computing … Manna for Analytics ?

There have been many blogs and articles written about cloud computing .. but then what is meant by cloud computing and why is it all the rage right now ?

Cloud computing :
This is a scenario where everything right from hardware to the software is hosted on someone else’s system and you use the same on demand…
This would mean that ( ideally speaking ) no more support pack upgrades , no more memory issues etc – just pay for what you need… which opens up a lot of possibilities …

Cloud computing and Business Intelligence

What does this do to the world of analytics….

Analytics in the SAP space have been largely restricted to :
1. Custom xApps which run specific analytics
2. Analytics that come as part of SAP BI ( Largely ignored due to lack of proper direction and not being packaged appropriately )
3. Standard Analytics that get delivered as part of modular implementations ( many of them are available but not used due to lack of knowledge about the same )

So what does analytics in an SAP BI world really translate into ?
This usually means dumping out the data required into large flat files / SQL tables using available interfaces … then passing the same through custom domain specific models built on either SPSS or SAS etc and then validating the results. This data is initially used to train the data model and then certain adjustments are made and the model is finalized and the data run through it.

The above process takes about 2 quarters to complete considering the amount of data to be handled and the model complexities.
This means that by the time the analysis is done the information is old … and this maybe is one reason why many people do not attempt data mining on a data warehouse but leave it to the analytics folks who just tell them what data they need.

Also coupled with this is the cleanliness issue that the results you get out of your system is as clean as your system’s master data is…

Now imagine that you have one single system that can host an SAP BI system as well as have a SAS system running. You just pay for the services alone. Now the database may be the same and you load your data mining model into the SAS instance and point it to the relevant tables in your EDW and you are up and running…. interesting and saves you a lot of time.

Also think about adhoc analysis of data – no more considerations of disk space for data marts – build the same – evaluate the data and then remove the same if not required..

How does this help us ?
1. No installation required – no more manuals to read and figure out what to install
2. Payment is based on the services used which gives a lot of room for optimization
3. No hardware cost ( at least the initial capex is avoided )
4. No more maintenance crews for 24×7 uptime of your servers etc…

All of this seems too good to be true in the real world but then people have started taking a stab at this .. the first being Amazon with its Elastic Cloud Compute or EC2 and many partners joining hands for the same.

We might think … will SAP be a part of the same but then the technology is too new and it takes a leap of faith in the mindsets of the corporate world to adopt the same and see economies of scale for this to be viable in the long term and not become something like the famous network pc….

Links :
Amazon’s Elastic Cloud Compute

Cloud Computing at Business Objects

Cloud computing and SOA

 Much of the information written about above might already be available in pieces all over the place but then everytime I did a search on BI trends for 2009 , the forst term was cloud computing .. thought I should explore the possibilities of the same in an SAP Landscape …

You must be Logged on to comment or reply to a post.
  • If your readers are interested in SAP and analytics on demand, they should take a look at the SAP BusinessObjects Business Intelligence onDemand solution (

    SAP BusinessObjects was one of the pioneers of onDemand solutions. We can host your data warehouse, cleanse your data, and handle your reporting for a per-user monthly fee, without you having to invest in any servers.

    And there’s also an “information ondemand” store ( where you can augment your corporate information with data from sources like Dunn & Bradstree, Thomson Financial, etc. — think of it as “iTunes for data”.

  • I think we should limit our expectations on cloud computing to the hardware benefits solely.  Instead of paying for and maintaining our own hardware we can let somebody like Amazon do this for us.  All we need to do is install and maintain the software on it.

    That said, it doesn’t help any integration issues at all between the systems, but it does allow potential selling / distribution of data that was typically harder to do before (IE: turn your data into a service and sell access to the data).