Capital One Software, the software arm of Capital One Financial Corp., today is announcing new features in Slingshot, which is a set of templates, custom workflows and controls that help users of the Snowflake Inc. Data Cloud better manage their cloud compute costs.

The offering is the first, and so far only, product the company has released in the year since it debuted, but more are expected as the unit gains its sea legs in the industry and builds a customer base.

“The last year for us has been about making sure that we’re delivering the right product for the right customer,” said Patrick Barch, senior director of product management. “We want to make sure it’s the right product for a broad set of customers and not for financial services or life sciences or other regulated entities.”

Cost visibility

Slingshot provides transparency into the cost, performance and usage of a company’s Snowflake accounts across variables like storage, serverless tasks, query timeouts and cloud services. Filters and tags can be applied to break down warehouse costs by custom categories.

The software continually scans the Snowflake environment to identify imbalances such as over- and under-provisioning. A dashboard shows the areas of greatest potential savings and provides targeted recommendations for improvement.

“We’ve created a way to dynamically provision compute capacity based on your specific peaks and valleys of usage,” Barch said. “Our recommended schedule may find that you don’t have a ton of activity between 12 a.m. and 8 a.m. so you can shrink your warehouse size to four clusters whereas during the workday you should scale to five clusters.”

A projections capability shows spending patterns over time, “which is really helpful when you go to renegotiate your contract or plan to add increased projects,” he said.

Identifying poorly performing warehouses and tweaking configurations is a bit of a black art, Barch said. “The first time a shared services team will hear that there’s an issue is when somebody from the business side is complaining,” he said. Slingshot gives data engineering teams a more proactive understanding of problem areas before they become apparent to users.

Better queries

A major enhancement in the new version is Query Advisor, which uses patented algorithms to dispense advice on how to improve SQL queries by comparing them to a database of thousands of optimized examples. “We scan the query text, run it against a set of optimization best practices and through an algorithm that comes up with recommendations that help you improve performance,” Barch said. “It’s something we developed in-house over the past several years and it’s helped us reduce our cost per query by about 43%.”

Suggested queries are segmented by opportunities for improvement and the recommendations are documented so users can choose which to implement.

The new version also provides greater visibility into resource usage. “You can now get a much deeper picture of your spend across all Snowflake cost drivers, whether it’s compute, storage, serverless tasks or cloud services,” Barch said. “Customers can better understand their bill, where they’re spending their money, and how to optimize what they’re doing.”

This version also includes improved tagging that makes it easier for users to tailor data sets to their needs.

“If you’re a retail company, for example, and you want to track shrinkage across different brands with out-of-the-box Snowflake, you would need to go into each individual account, manually apply the same set of tags, build a custom process to export all of that data into a central location and stand up a [business intelligence] tool on top of that to track one common process outcome across all of your different accounts,” Barch said. “With our new release, you can do that all in one step from a single pane of glass across all of your accounts.”

Image: Capital One Software