The Cloud Is Great for Data, Except for Those Super High Costs

The cloud has revolutionized the fields of advanced analytics and AI, as well as IT as a whole. Companies now have access to practically limitless storage and compute resources for crunching massive datasets and training machine learning models. However, the unexpectedly high costs of cloud computing increasingly is becoming a problem.

In a February report, Gartner predicted that total spending on the cloud will grow to $544 billion this year, a 21% increase from last year. While it still trails spending in on-prem systems ($775 billion this year), cloud spending is projected to exceed on-prem spending by 2025, when Gartner predicts $917 billion will be spent on the cloud.

Clearly, the cloud giants are gobbling up vast quantities of workloads. The big three public cloud companies–AWS, Google Cloud, and Microsoft Azure–are three of the five biggest companies in the world, with a collective market capitalization in excess of $4.5 trillion. (Only Apple and Saudi Aramco, with another $4.5 trillion between them, are bigger.)

Cloud providers aren’t the only ones who are benefiting from this massive migration to the shared computing model. During COVID-19, companies accelerated the retirement of on-prem data centers, thereby maintaining social distance requirements.

But beyond that, the availability of cutting-edge technology in the cloud is also helping customers get a good return on their cloud investments. A recent Deloitte survey found that 88% of IT decision makers believe that the cloud is the “cornerstone of [their] digital strategy.”

One company that has gone all-in on cloud is Capital One, the credit card company based in Virginnia. Capital One has been moving to the cloud for years, and it recently completed the project when it migrated an on-prem Teradata data warehouse to a Snowflake warehouse running on AWS.

(Hermin/Shutterstock)

While the move from Teradata to Snowflake finally gave thousands of Capital One users the capacity to run running millions of queries, the newfound scalability came with some unpredicted impacts, according to Salim Syed, a vice president and head of engineering at Capital One Software.

“But the [consequence] with that kind of unlimited power and unlimited compute is you can go from data starved to data drunk very easily,” Syed recently told Datanami. “You can end up blowing through all your credits if you don’t have proper governance, proper cost control measures in the way you’re provisioning your data platforms.”

Capital One searched the market for potential data management solutions for its Snowflake environment and, dissatisfied, decided to build its own, which it now sells through Capital One Software through the Slingshot brand.

Capital One isn’t the only enterprise with that sort of experience. A recent report from Gartner found that companies’ cloud bills are frequently two to three times higher than expected. What’s more, it found that up to 80% of companies consistently go over budget on their IaaS spending.

Stephen Brobst, the CTO of Teradata, cited another analyst study that concluded that 80% of organizations deploying data warehousing technology go over budget by more than 50% in the first 18 months of deploying in the cloud. “This is a very unpleasant discussion to have with your CFO,” he said.

While resources such as CPU, I/O, and storage essentially are infinite in the cloud, that’s not an excuse to stop managing the consumption of resources, which Brobst accuses competing data warehousing vendors of doing.

(sdecoret/Shutterstock)

“From a financial point of view, that’s unacceptable and irresponsible,” Brobst told Datanami. “We believe that it’s important to have the elasticity, but to have a FinOps approach–that’s Gartner’s term–to manage the resources.”

While it continues to maintain its customers’ on-prem warehouses, Teradata has shifted its strategy to the cloud, which now accounts for the lion’s share of new customer revenue. Its main competitor is Snowflake, which is the target of the new cloud data lake offering that Teradata unveiled last week.

Another tech firm taking a bite out of the cloud-cost apple is Anodot, which develops machine learning and AI-based business monitoring software. According to its just-released 2022 Stat4e of Cloud Cost Report, 49% of businesses surveyed “find it difficult to get cloud costs under control, and 54% believe their primary source of cloud waste is a lack of visibility into cloud usage.”

Other findings from the report include:

Over a third of participants (37%) have been surprised by their cloud costs or had an incident related to cloud costs;
More than half (53%) say the key challenge is gaining true visibility into cloud usage and costs, while 50% said complex cloud pricing and 49% said complex, multi-cloud environments;
More than one quarter (28%) of respondents said it takes weeks or months to notice a spike in cloud costs, a figure that has not improved over 2021.

Customers moving to the cloud should leverage AI and machine learning to help monitor their cloud resources, according to Anodot CEO and co-founder David Drai.

“Shifting to the cloud requires a delicate balance between the speed of workload migration and cost control,” Drai says in a press release. “Today, cloud cost management – which is all the more crucial as businesses strive to mitigate wasted resources and shore up revenues – should be based on four key elements: visibility, insights, recommendations, and automated actions. Enterprises that are leveraging AI and ML to control and optimize their cloud environments are seeing immediate results and reduced waste, unlocking the true value the cloud has always promised.”

Cloud Migrations Negatively Impacting Data Estates, Capital One Says

Cloud Getting Expensive? That’s By Design, But Don’t Blame the Clouds

The post The Cloud Is Great for Data, Except for Those Super High Costs appeared first on Datanami.