Just over a year old, digital-content-archiving vendor Sonian Inc.'s infrastructure is almost entirely cloud-based. No in-house servers, data centers or storage. So when its data warehouse provider, Andover, Mass.-based Vertica Systems, made its analytic database available on-demand earlier this month, Sonian executives didn't think twice, quickly trading in their internal Vertica data warehouse for the cloud-based version.
"We're a cloud-based archiving solution," said Greg Arnette, Sonian's founder and chief technology officer. "To have Vertica in the cloud, where it's a managed service that we can subscribe to ourselves in the same way that we subscribe to compute and storage, it was really a winning combination."
Sonian, based in Dedham, Mass., provides hosted email and instant message archiving for its customers. After customer metadata is loaded into Vertica's on-demand data warehouse, which is hosted by Amazon's Elastic Compute Cloud (EC2), Sonian makes the data available to its customers for analysis via Web-based services. EC2 is part of Seattle-based Amazon's Web services business in which customers like Vertica rent server capacity from Amazon, i.e., the cloud, to power applications, rather than using internal servers. Customers can rent more or less server capacity as demand fluctuates.
Proponents of on-demand business intelligence (BI) and data warehousing argue that the cloud-computing model lets businesses access and analyze data more cheaply and quickly than internal BI and data warehouse systems do. Further, maintenance and upgrades to cloud-based analytic databases are handled by host vendors, allowing customers to direct their IT resources elsewhere.
Vertica Analytic Database for the Cloud, for example, lets users create and access analytic data warehouses on-demand over the Internet in as little as 30 minutes, the company says, avoiding data center costs and the delays that come with in-house analytic database deployments. The SQL-based database boasts a column-oriented architecture, which Vertica and industry experts agree is better suited for analytic functions than row-based databases are.
Customers pay as they go, purchasing additional server power as workload demand increases. "There's no provisioning of hardware or configuration needed," Omer Trajman, director of field engineering for Vertica, said in a demonstration webinar. "And then when you're done with the system, you shut it down and stop paying for it."
Vertica's on-demand analytic database starts at $2,100 per month for up to 500 GB of data.
Market for on-demand data analytics limited
Still, analysts say the market for on-demand data analytics is limited. Boris Evelson, a principal analyst with Cambridge, Mass.-based Forrester Research, said large enterprises with the capital, infrastructure and internal talent to deploy and maintain in-house BI applications and data warehouses aren't likely to adopt the on-demand model.
And very small companies, those running just a single application suite like NetSuite or QuickBooks, can probably get all the BI and data analytic capabilities they need from that application suite, Evelson said. That leaves just midsized companies as potential on-demand BI and data warehousing customers -- those companies "that have outgrown the single IT environment but are not big enough to have an IT staff that's sophisticated enough to know how to implement business intelligence," he said.
Even for midsized companies, BI and data warehousing, whether in-house or on-demand, still require plenty of heavy lifting, further limiting the appeal of cloud-based data analytics.
"BI is still an art much more than a science," Evelson said. "It still takes an army of consultants to pull it together, and whether I'm hosting it somewhere in the cloud or doing it in-house, I'm still going to go through exactly the same difficulties. And as long as I'm doing that, why would I want to release -- or lose control over -- my BI installation to a third-party vendor?"
One exception in large enterprises, he said, is the so-called power user who needs to slice and dice data quickly in ways not supported by internal BI systems. An on-demand data warehouse for such analysis that can be quickly and cheaply deployed without a long-term commitment, and easily shut down when it's no longer needed, might suit their purposes better than deploying an internal data warehouse through IT, which takes time and coordination.
Companies like Sonian, which operate entirely in the cloud and are comfortable with hosted computing, are another potential customer base. They are more likely to embrace on-demand BI and data warehousing than those companies with a traditional IT infrastructure, according to Philip Howard, research director with U.K.-based Bloor Research. Of course, the list of companies that rely solely on a cloud-based architecture is a short one.
Speed, security questions dog cloud-based data warehousing
Speed and security are two more potential drawbacks to cloud-based data analytics. Both Evelson and Howard said that any on-demand BI system or data warehouse can load and analyze data only as fast as the Internet connection used to access it allows.
"I know [Vertica] talked about having a fast loader, but you're still loading across the Internet, so how fast can it be? It's a possible constraint," Howard said.
Evelson and Howard also question how willing companies are to let their data outside the safe confines of corporate firewalls and float out in the cloud. Data used for business intelligence purposes is among the most sensitive that companies collect.
"BI today is one of the main mission-critical applications. You typically put applications in the Software as a Service cloud when they become a commodity," Evelson said. "That's not the case for BI. It's very highly proprietary data. You absolutely want to ensure that nobody's going to see it, so you're probably not going to put it out there into somebody else's hands."
Still, for midsized companies without the resources to support an in-house BI system or data warehouse, taking a chance on the cloud-based model might be worth the risk, Evelson said, especially if the only alternative is to let data sit idle in a database.
Certainly, that's what Vertica and other BI and data warehousing vendors with hosted offerings -- including 1010data and Kognitio -- are banking on. Of course, they're not giving up on the traditional data warehousing deployment model either.
Does your company use on-demand business intelligence? Email SearchDataManagement.com editors with your story!