Ahana Cloud Inc., which curates a commercial version of the Presto open-source distributed query engine, today is announcing numerous updates to its recently introduced Presto managed service.

Topping the list is an input/output caching capability that executives said increases performance up to five-fold. Based on the RubiX open-source lightweight data caching framework, the feature eliminates the need to read data from data lakes, particularly if the same data is read over and over. Enabled with a single click, the service automatically attaches solid-state drives and sizes the cache at the container level based on the user selected instance type.

“The cluster uses each cache on a per-user basis for up to 85% reduction in processing time for needle-in-a-haystack type of queries,” said Dipti Borkar, co-founder and chief product officer.

New clusters now come pre-configured with additional advanced features such as memory tuning, cost-based optimization and dynamic filtering capabilities for better performance on complex queries. Idle clusters can be configured to be shrunk but not shut down so that they can be restored if a query comes in, Borkar said. “If you don’t need the resources you shouldn’t have to use them,” she said.

New support for asynchronous query execution using Apache Superset enables long-running queries to proceed without tying up administrative infrastructure. Ahana manages and automates the deployment of the required infrastructure for these queries to run asynchronously.

“If a query runs for 30 minutes your administrative console is stuck,” Borkar said. “This enables you to go on with your work on the admin panel. The value is efficiency for the platform engineer.”

In the area of security, integration with the Apache Ranger framework permits fine-grained access controls to be applied to Presto clusters.  “It allows users to define policies on who has access to what,” Borkar said “Presto didn’t have something to do this that was open source.” Ahana will contribute the Ranger integration code back to the Presto project.

Policies needed for Ahana management of the compute plane have been tightened and made more transparent through tags, effectively isolating the Ahana environment from any other user-managed resources.

Image: Pixabay