Data Ops Upgrade Targets App Performance
The field of data operations is undergoing a shift toward automated tools designed to boost application performance and, with it, productivity, as data sets grow.
With that in mind, “data ops” startup Unravel Data this week unveiled the latest version of its application performance management platform that automates data operations tasks ranging from problem discovery to root-cause analysis and resolution.
The company, based in Menlo Park, Calif., defines data ops as a set of practices and tools used to accelerate data analytics while improving the reliability and quality of insights. “Done right, DataOps fosters a tight collaboration between data engineers, data scientists and IT operations,” the company asserts.
IT teams typically use a collection of tools to manage data stacks, including application logs and control systems used to administer a cluster. Unravel Data argues that approach creates bottlenecks and overburdens harried data ops teams.
The startup’s approach as embodied in it 4.0 release is a “full-stack” software platform that integrates the monitoring of applications, clusters and underlying infrastructure. Among the advertised benefits are identifying bottlenecks in running Spark, Hadoop and other workloads along with providing a window into how resources are being used on a cluster. That in turn is designed to help users predict computing requirements so they can scale cloud infrastructure.
The results, the company claims, are more reliable applications and improved cluster utilization.
As data-driven enterprise infrastructure is increasingly used to deliver distributed applications, the upgraded management platform tracks emerging technologies such as application containers to ensure they are fully utilized. At the same time, it monitors applications to determine whether computing and storage resources are being misallocated. If detected, the management tool is designed to reallocate those resources to optimize resources for an application.
Other upgraded features include alerts about bad configurations in a Hadoop and Spark clusters along with the application of new configuration settings. The platform also provides alerts if storage limits are being reached. The storage utilization and caching feature identifies which tables and files can be removed or cached to free up storage.
Unravel Data, which emerged from stealth mode last September, said its 4.0 application management platform is available now for on-premise, cloud and hybrid deployments. The company said early customers such as Autodesk (NASDAQ: ADSK) and YP.com (formerly, the “Yellow Pages”) are currently using the system to manage their production workloads.
“End users are spending more than half of their day trying to solve these issues and getting productive on the big data stack,” Kunal Agarwal, the co-founder and CEO of Unravel Data, told Datanami last year. “People don’t necessarily understand distributed computing and parallel processing, and getting performance and reliability out of these applications is super hard.”
The startup’s founding team includes veterans from MapR, Microsoft (NASDAQ: MSFT) and Oracle (NYSE: ORCL).