Multi-cloud strategy becoming a necessity for BI vendors

The exceptions

Despite the flexibility a multi-cloud analytics environment offers customers, a vendor may have reasons choose to support just one cloud as opposed all of them.

They're specialized cases, but not all BI platforms are designed to appeal to a mass audience.

If a vendor targets a specific type of customer, and tailors its platform to one particular industry, one cloud might be a better fit than another, according to Joe Hellerstein, founder and CSO of Trifacta, a data wrangling vendor.

And just as that could mean one cloud is a better fit, it could also mean another is not.

"Because Amazon is such a huge retail player, many retail customers don't like to run on AWS -- they would prefer to run on Microsoft or Google -- so tech companies that are targeting the retail space might have a disincentive to work on AWS," Hellerstein said.

Another reason is the added cost and complexity that goes into BI vendors creating a multi-cloud analytics environment, according to Hellerstein.

"Focus," he said. "Unless you're a very large vendor, focus is always good, so it's challenging to be cross-platform -- it costs in engineering, it costs in marketing, it costs in closing deals. But the payoff for most software vendors outweigh the costs."

And the payoff is being able to serve a broad customer base.

Vendors who don't offer multi-cloud environments are somewhat inflexible, and that kind of inflexibility serves as an ultimatum of sorts. It doesn't matter to the same degree if a vendor is only targeting a small customer base, and group of organizations all in the arts, or sports, or an emerging industry such as cannabis.

BI products need to adapt to their host environment. So if the customer's architecture is multi-cloud that means the BI product needs to be multi-cloud. It needs to pull from applications running on multiple clouds as well as on-premises systems.

But to appeal to a wider audience, enabling flexibility is critical.

"BI products need to adapt to their host environment," said Wayne Eckerson, founder and president of Eckerson Group. "So if the customer's architecture is multi-cloud, that means the BI product needs to be multi-cloud. It needs to pull from applications running on multiple clouds as well as on-premises systems."

Cloud differences

If all clouds were the same, or if one cloud service provider held a dominant market position and the others were only lesser players, there wouldn't be the same need to support multiple clouds.

But the clouds -- whether from AWS, Microsoft or Google, the cloud providers with the biggest market shares -- aren't identical.

Organizations, therefore, are often multi-cloud in their own right and therefore need a BI platform that offers a multi-cloud analytics environment.

Multinational organizations, in particular, need to use more than one cloud, according to Tibco's Quinn. Not all clouds are available in all regions of the world, and even if they are, clouds, like anything else, sometimes have technical glitches and organizations need to keep their operations running while cloud providers fix those problems.

"Multinational companies had their cloud vendor of choice but while it was available in some locations, it might not be available in another, so they use another cloud," Quinn said. "Then they started looking at multi-cloud to protect their investment -- they have a preferred platform, but they use another in case there's an outage at their preferred platform."

Similarly, MicroStrategy's Saylor -- whose company provides BI services to numerous U.S. government agencies -- noted that in order to do business overseas it's critical to have a multi-cloud strategy.

"You can't do business with certain governments without being able to deploy in their secure facilities -- you're just out of business," he said. "And then, on the other hand, you can't do business without being able to deploy in AWS for a certain group of companies, but another group of companies would absolutely disqualify you from doing business if you have them deploy on AWS -- they would insist on Azure."

Beyond location, the various clouds differ in their compute capabilities. Those compute differences -- better Augmented intelligence capabilities, for example -- can make a certain cloud more appealing to one organization and another more appealing to another organization.

And BI vendors naturally want to appeal to as many customers as possible.

"At a general level, different cloud vendors have different strengths," Quinn said. "Google is good for data and AI tools, Amazon has a lot of services. If you explore them, you can see the slight differences. If you look at pure compute storage, all three work."

I also see customers choose multiple clouds for projects to handle different parts -- they're trying to use the unique parts of the platform," he continued.

Fear of vendor lock-in

Vendor lock-in has long been a concern for organizations.

Some, who first got started with what are now often referred to as legacy BI vendors, have built up such complex data stacks with that one vendor over the years that they're essentially trapped. To transform their analytics operation would be complicated and expensive, and so they remain locked in with the vendor they chose perhaps a couple of decades ago.

It's why many organizations now use different vendors for different parts of the analytics process.

Even if a vendor offers everything from data collection through storage, preparation and visualization, enterprises may choose one of the cloud service providers for data storage, Alteryx or Trifacta for preparation, and Tableau or Qlik for visualization. Maybe they even use Narrative Science for data storytelling.

With Microsoft offering tools that span the analytics process, Google acquiring Looker last spring, and perhaps AWS in the market for a BI vendor of its own, vendor lock-in is now a fear amid the migration of analytics to the cloud.

By storing their data in different clouds, however, organizations preserve some autonomy.

"The evolution toward companies wanting to be multi-cloud also has to do with minimizing vendor lock-in -- by choosing a multi-cloud strategy and going in with a multi-cloud position, it forces the architectural and technical decisions that you can't make if you're only using one vendor," Quinn said.

Similarly, Saylor noted that customers often fear having to deal with just one vendor and want flexibility as a defense against vendor lock-in.

"I never met a customer who insisted on Azure that said, 'And I'm comfortable having only Azure and no other option,'" he said. "They say they don't want to be locked into Azure and they want to try Google too."

Flexibility, meanwhile, means not just supporting a multi-cloud analytics environment but also supporting those customers who haven't yet migrated to the cloud and still run their analytics operations on premises, AVOA's Crawford said.

For some organizations -- generally small and midsize business (SMBs) -- there's simply no need to migrate to the cloud because they generate their data on premises, store it on premises and analyze it on premises. For others, again mostly SMBs, cloud costs are prohibitive. And for still others, the complexity of migrating an entire operation to a new platform is daunting.

"If a customer has their data on premises and still generates their data on premises, they may prefer to lean toward an on-premises solution," Crawford said.

Flexibility, ultimately, is key for BI vendors trying to meet customer needs. And a multi-cloud environment, ultimately, is about giving customers flexibility.

"Generally, the customers want to keep their options open," Saylor said. "And keeping their options open means migrating, on premises, off premises, between various clouds, and the most friction-free fashion."