Pros and Cons of Business Intelligence in the Cloud

Ann All

Last week I wrote about a teaming of four vendors to offer a cloud-based stack of business intelligence software. It sounded like an intriguing proposition, especially for smaller companies whose BI budgets (if they had them at all) lagged those of their larger counterparts.

 

These vendors aren't the first to make a BI play in the cloud. Amazon is among those that came before them, offering a solution in which customers can build SQL Server and Oracle databases in its Elastic Compute Cloud. Relational databases appear to be one of the hottest cloud applications around, as IT Business Edge blogger Art Cole wrote earlier this month.

 

Sounds exciting, right? Yet as is often the case with breathless product announcements, there aren't a lot of specific details, like how and why companies might want to employ BI in the cloud. For that, I found a Beye Network article in which two guys from Persistent Software, Mukund Deshpande and Shreekanth Joshi, concisely lay out the benefits, challenges and approaches of moving BI to the cloud.

 

As always with content penned by vendors, keep in mind these guys are responsible for the BI and software-as-a-service businesses, respectively, for Persistent and would surely like to sell you something. They don't try to do so in the informative article, however. The key reasons they cite for using BI in the cloud:

 

  • Predictive analytics and other BI applications can gobble lots of computational resources. So it makes sense to do them in the cloud, where you can scale those resources as you need them. (The cloud's ability to scale for these kinds of apps was also cited by several sources I interviewed for a story on cloud computing, In the Cloud, Process and Budget Questions More Thorny than Tech. In particular, why make that kind of an on-premise infrastructure investment for apps you may use only sporadically?)
  • You can store heavy data loads for less in the cloud.
  • The cloud makes it simpler to bring reporting and visualization capabilities to more users, more quickly.
  • And of course, there's that old chestnut, that it's easier and quicker to attain ROI in the cloud. (Keep in mind that not everyone agrees the cloud is always more cost effective.)

 


The key challenges:

 

  • Moving large data sets to the cloud could get costly. They recommend shipping disks, an approach they say is often recommended by cloud providers like Amazon.
  • Though encryption keys, SSL and certificates can be employed to store data securely in the cloud, some data may be so sensitive that it needs to remain on-premise. (Security and compliance issues are a continuing challenge. As IT Business Edge blogger Lora Bentley wrote earlier this week, Amazon Web Services can't offer Level 1 compliance with the PCI Data Security Standard.)
  • Established BI vendors thus far are only offering limited services in the cloud. (Guess that's why the BI stack announcement got so much attention.)
  • Integrating on-premise data with cloud components is still a challenge. (Cloud integration issues will linger for a while, wrote IT Business Edge blogger Loraine Lawson.)

 

Deshpande and Joshi also make some interesting predictions for the future. Among them:

 

  • Some vendors will embed BI services within their platform-as-a-service (PaaS) solutions.
  • More public data will become available on the cloud, which will drive the further development of cloud-based BI solutions.


Add Comment      Leave a comment on this blog post
Aug 20, 2009 4:34 AM Fiona Fiona  says:

I would go on a system that is creating the clouds inside the organization

There are a few solutions that are doing just that and in a very competitive price comparing to the cloud offering.

Reply
Aug 24, 2009 1:34 AM Dan Graham Dan Graham  says:

There are many pros and cons of putting BI in the cloud.   Some workloads are perfect for clouds, others would be insane.  You first have to segment the word cloud into public and internal private clouds. 

 

Public clouds are great for numerous workloads where the cloud is 'good enough'.  This often means the security or performance needs are not mission critical.  It turns out there are hundreds of spread marts and data marts that fit in this category.  So a public cloud is a good place for proof of concept testing, short bursts of quality testing, developers, and even some small data marts.  Performance is 'good enough' and the prices are often attractive.  But public clouds have not yet offered prices for performance goals or service levels for availability or security.  These vendors aren't dumb-they are diligently working on it.  But the offers don't exist yet.  So occasional use or basic BI works best in public clouds for now. 

Internal private clouds are the alternative, often built on virtualization software or with interfaces to the existing IT managed data warehouse.  Because the private cloud is inside the firewall and managed by the IT Operations staff, performance, security, and availability can be managed to mission critical levels.  So some companies are adding cloud capabilities to their existing data warehouse.  Self service, on-demand provisioning, multi-tenant, and dynamic capacity are achieved on the 'box' in your data center.  So these configurations work well for mission critical data warehouses and of course data marts, predictive analytics, etc.  The down side is you still have up front capital costs. 

There is a lot to learn if it's your first time using a data mart in a public cloud.  It's not going to be easier, simpler, or give a fast ROI your first couple tries.  Those things come after you have solid experience with the clouds.  Downstream, we will all learn more about which applications should go into the clouds, which shouldn't.  Like outsourcing, the rule of thumb is only put applications in the cloud that are not strategic core competencies of your organization.  Often this includes mission critical applications.  BI/EDW will be the same.

I recommend getting started on projects that don't carry a lot of risk.  Learn, let the vendors mature, then learn some more.  This puts you ahead of competition and when clouds mature each year, you stay ahead.

Reply

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

null
null

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.