RESEARCH & RESOURCES

Tableau Takes on the Cloud

Tableau says its new Tableau Cloud offering gives users the best of both worlds.

There's wasn't much said about cloud at last month's Pacific Northwest BI Summit, held in Grant's Pass, Ore., despite the fact that all attendees represented vendors that have cloud strategies.

What's more, at least one -- Australian business intelligence (BI) import Yellowfin, which was represented by its CEO, Glen Rabie -- is a cloud-only proposition.

Nevertheless, cloud just wasn't a big topic of discussion. One vendor that was keen to talk cloud was Tableau Software Inc..

This isn't surprising: last month, Tableau officially launched its long-incubating Tableau Cloud solution. The company had been quietly selling Tableau Cloud for months: in a May interview, for example, Ellie Fields, senior director of product marketing with Tableau, disclosed several Tableau Cloud pilot projects -- including a hedge fund start-up that expected to run its entire BI and analytic practice on Tableau Cloud.

"The founders were veterans from financial services. They wanted to start a hedge fund; they didn't want to start an IT team building out IT systems," she explains. "For them, Tableau-in-the-cloud fit very nicely with what they wanted to do. If they have five people on [Tableau], and then 20 people need it, all they have to do is just add new users. They not only don't need an IT team to go and do a big purchase decision, they also don't need them to go and provision new servers and deploy them."

Tableau Cloud isn't a completely off-premises design. Would-be users must first provision at least one on-premises Tableau system -- be it Tableau Desktop or Tableau Server -- to start using it. The reason, explains Suzanne Hoffman, senior director for analyst relations with Tableau, is that Workbooks and other Tableau artifacts can only be created in an on-premises Tableau system. Once users or analysts create a Workbook, they can expose it in the cloud.

"If you look at who uses the cloud, it tends to be smaller companies -- people who don't have IT support. So what we did is rather than starting with ten users, which is what Tableau Server starts with, you could have one person ... [creating Tableau artifacts] and all of a sudden you have the capability of sharing Tableau in the cloud with three people, with five people. You can add incrementally as many people as you want."

Fields and Hoffman describe a couple of scenarios in which Tableau Cloud might have legitimate sex appeal. The first is as an analytic platform for other cloud services -- software-as-a-service (SaaS), platform-as-a-service (PaaS), or infrastructure-as-a-service (IaaS) offerings from familiar vendors (Salesforce.com, Google Analytics) and cutting-edge offerings (Amazon's Redshift data-warehouse-as-a-service, Google's BigQuery). In this use case, Tableau Cloud is largely consuming live feeds -- using REST APIs -- from other cloud apps.

"These are the apps which generate cloud data, the Salesforce or Google Analytics apps: apps that live in the cloud. You don't want [Tableau Cloud] to do a live connect to [on-premises] transactional systems, because that's going to impair [the performance] of the system. You're more likely going to do [a scheduled] extract," Hoffman explains.

"Your [on-premises] transactional applications can still be a source of data, it's just that as opposed to Amazon RedShift or TreasureData [a start-up big-data-in-the-cloud service], you're just doing a cloud extract. With Amazon Web Services [AWS], you're doing a live connection back and forth between Tableau [Cloud] and data that lives in Redshift."

Another potential use case involves tapping Tableau Cloud to complement an existing on-premises Tableau Server practice. The idea is to expose Tableau Cloud to traveling, telecommuting, or mobile users -- chiefly, to sales people.

"There can be an advantage to having both traditional BI and [a] cloud BI [service] outside the firewall. People on the road don't want to have to go through your firewall or VPN," Fields told BI This Week. "They like the ability of authenticating on their devices and going straight into the cloud without taking that additional step of establishing a VPN [connection]," she continued, noting that connecting to a VPN is a frequent source of frustration for users.

"A secure cloud solution means that your people on the road won't have to spend an extra five minutes in frustration trying to negotiate your VPN."

Cloudy with a Chance of Obsolescence?

This last use case involves a kind of hybrid cloud deployment -- although what Tableau means by a "hybrid" cloud isn't what (for example) cloud specialists such as Eucalyptus Technologies Inc. (which develops an AWS-compliant private PaaS offering), Red Hat Inc. (which develops the OpenShift and OpenShift Enterprise PaaS cloud stack), or ActiveState Software Inc. (which develops the Stackato PaaS cloud stack) might mean.

In the latter cases, a "hybrid" cloud describes simultaneous on-premises (i.e., private) and off-premises (public) PaaS deployments: applications that can, notionally, be shifted back and forth between private and public PaaS contexts. PaaS itself is a vision of an elastic platform that can be deployed either internally or externally -- and on stacks that support it, both internally and externally -- and which exposes a storefront full of pre-packaged applications that can be downloaded and installed (but not necessarily configured and deployed) with a few clicks of a button.

Tableau Cloud, on the other hand, is a fit-for-purpose application: it's software-as-a-service (SaaS) in an almost classic sense. In this respect, it's similar to cloud offerings from other traditional (that is, on-premises) BI vendors, which are by and large SaaS-oriented.

That said, tools from Actuate Corp. (the BI Reporting Tool), JasperSoft Inc., and Talend are available as PaaS packages; Microsoft BI is available via Azure, which doubles as both a PaaS and as an infrastructure-as-a-service (IaaS) stack; Yellowfin, for its part, says it supports Yellowfin BI on AWS, Azure, and other PaaS or IaaS stacks.

Tableau Server, in fact, can be deployed on AWS and Azure: prior to the introduction of Tableau Cloud, several Tableau customers did precisely this, typically in combination with Amazon's Elastic Compute Cloud (EC2). This isn't the same thing as a PaaS-ready version of Tableau.

It isn't clear that such an offering is needed or wanted at this point, however. In spite of the undeniable appeal of offerings such as AWS -- to say nothing of the excitement stirred up by offerings such as Redshift -- PaaS might still be an idea ahead of its time. At this point, argues Mark Madsen, research analyst with IT strategy consultancy Third Nature Inc., there just isn't much of a market for BI PaaS. Even though an ROI case can be made for PaaS on the basis of projected hardware or operational savings, it's still a tough sell to management. SaaaS, Madsen argues, is a much easier sell. He cites MicroStrategy Corp.'s cloud strategy as an example of a BI vendor having struck a pragmatic balance, with distinct PaaS and SaaS cloud offerings.

Tableau seems to have gotten it right, too. By at least one important metric, it's given its customers exactly what they want in Tableau Cloud: in less than two weeks of public availability, officials say, Tableau has already logged well over 1,000 Tableau Cloud "try-outs."

TDWI Membership

Get immediate access to training discounts, video library, BI Teams, Skills, Budget Report, and more

Individual, Student, & Team memberships available.