Cloud Analytics Engines Falling Short, Survey Finds
There’s an unmistakable shift underway toward cloud analytics, but a vendor survey found an analytic engine “performance gap” when it comes to running cloud workloads at scale.
The performance shortfall that grows exponentially for the largest enterprises stands in contrast to the industry consensus that the cloud is the ideal platform for analytics. Fully 83 percent of the 700 senior executives polled by analytics platform specialist Teradata Corp. (NYSE: TDC) said the public cloud was the best place to run analytics workloads. A higher percentage said the transition should be happening faster.
While problems scaling analytics workloads in the cloud seem counterintuitive, a range of barriers to cloud analytics have emerged. Along with familiar concerns about public cloud security (cited by half of respondents), “immature” and underperforming cloud-based analytics engines were most often cited as hurdles to cloud analytics.
Other obstacles included stricter regulatory compliance, difficulty connecting legacy systems with cloud applications and a general “lack of trust” in cloud technology, which was cited by about one-third of respondents.
Part of the performance shortfall stems from the emergence of machine and deep learning tools used to deploy business intelligence and, increasingly, AI applications. The latter requires huge volumes of training data, taxing the capabilities of current analytics engines as they transition to the public cloud.
The Teradata survey released on Monday (April 23) found that one-third of respondents are using deep and machine learning to power AI applications. That total is expected to double over the next year.
Other leading cloud analytics applications include data visualization and data mining, with nearly three-quarters of companies polled saying they will shift these functions to the cloud over the next 12 months. The leading use case is customer service analytics followed by marketing and sales.
Still, cloud providers appear to be falling short in meeting high enterprise expectations for cloud analytics. Along with faster, flexible deployment, 44 percent said they expect the shift to cloud analytics to deliver better performance as well as “faster insight into data.”
Given those expectations, vendors like Teradata, Dayton, Ohio, are attempting to move beyond flagship databases to offer full-blown platforms that can help ease the transition to cloud analytics. Along with an analytics platform billed as the latest iteration of an enterprise data warehouse, the company also is targeting the cloud analytics transition with a cloud service aimed at the current performance gap identified in its industry survey.
“When it comes to deployment, the organizations surveyed are bullish on the cloud, but concerned about the slow pace of analytics adoption in the cloud,” the survey concludes. Teradata’s response is to offer a service for moving on-premise analytics infrastructure to the cloud to leverage scaling capabilities while delivering what it claims is its “industrial-strength analytics.” The cloud analytics service is said to handle 25 million queries per month.