A Report from the TDWI Solution Summit on Cloud BI in Scottsdale

We recently attended the TDWI Solution Summit on Cloud BI in Scottsdale, AZ.  A number of topics were covered, some of which may appear in future blogs but one of the more interesting panels focused on cloud BI myths.  Fern Halper from TDWI moderated the panel and raised two myths for discussion with a panel that included Suzann Hoffman from Tableau, Doug Schlemmer from Red Rock BI and Hannah Smalltree from Treasure Data.  The idea was to test each of these ideas against the experience of the panel and evaluate how much, if any, truth there was in the myth.

Myth #1.  The Cloud is not as secure as on-premise solutions.

Suzann argued that the sensitivity around Cloud based BI depends on the type of application.  She saw marketing and sales functions at the forefront of innovation with the finance and accounting function lagging behind in adoption.  Related to this point, she argued that governed data (e.g. compliance data, company financials etc.) often moves to the cloud last, partially due to the innate conservatism that companies have with regard to this type of data.

Doug pointed out that security is only as good as the policies behind it and not necessarily a feature of the on-premise and cloud channel.  He pointed out that for many companies, on-premise environments are not  secure and cited some examples we have all seen including:  open wi-fi, employees’ ability to access data on thumb drives, and the always surprising ‘server under the desk’.  In this sense, he argued the cloud often provides better security than on-premise solutions.

He went on to argue that the data represents the real security risk rather than the applications themselves.  So, he argued companies should focus on data in transit and data at rest as the key elements of their security efforts rather than focusing on the presentation layer, etc.  As a result, companies that want to manage sensitive data in the cloud should consider multi-tenant solutions for the presentation layer, but use a single tenant approach for the physical database itself.  This provides the best tradeoff between the cost benefits of multi-tenancy and the security of single-tenant solutions.

Hannah framed the issue a slightly different way, by asking the question, “Is your data center more secure than Amazon Web Services (AWS)?”  This argument resonated for me.  AWS has enormous scale and many, many customers.  They also have their own data security needs internally.  On the one hand, many companies that could use BI will likely never have the scale or resources to build an AWS-quality security framework.  The cloud could well be a better security option for these companies.  On the other hand, many have the resources, drive and expertise to develop security frameworks that work as well or better than AWS.  It feels like a case by case evaluation to me.  And so the debate will continue…

Myth #2.  It is more difficult to manage data in the cloud.

Fern highlighted the myth that companies find it harder to manage their data in the cloud for data transfer reasons.  In general, this felt like a more complex issue with a more nuanced set of answers vs. the security issue that sounded like a myth upon inspection.

The panelists raised several issues here.  Suzann discussed Tableau’s philosophy of limiting the number of times you need to poke a hole in the firewall.  Tableau attempts to limit the amount of data transferred back and forth to minimize the data transfer and latency issues.

Doug took a different point of view, arguing that the cloud now has very high data transfer rates.  Getting the data into the cloud does not need to take place via the internet.  For example, AWS offers physical options for data transfer.  In addition, with AWS Direct Connect, you can transfer data at up to 100 megabits per second.  So, usually he does not see latency inside the cloud.

Hannah made similar arguments, but highlighted the business need to align data transfer issues with the actual latency requirements.  If you need real time data, you will need to stream your data into the cloud.

One of the audience members challenged the panel based on the pricing model.  Transferring data into the cloud typically costs nothing (at least on AWS), but transferring data out can become very expensive.  As a result, managing data inside the cloud, the questioner argued, works better for applications where the company does not need to send data constantly between its on-premise and cloud environments.  This seemed to echo the comments from Suzann.

In general, the panelists seemed to agree that if the data already lives in the cloud, it makes it easier to manage BI in the cloud.  However, as Rikesh Bansal from Brocade pointed out in his presentation this morning, as more and more applications move to a SaaS format in the cloud, the need for companies to move data around the cloud and between the cloud and on-premises solutions will increase anyway.  Indeed, as more applications that companies use daily move into the cloud, it could become more difficult to manage that data via an on-premise approach.  In this sense, the myth, whether valid or not, will become increasingly irrelevant as a reason to stay out of the cloud and increasingly important as a general issue for companies to manage.