For Eliot Horowitz, CTO and cofounder of MongoDB, it’s been all about solving database problems. From his time as cofounder of a website called ShopWiki, where he developed crawling and data extraction algorithms, he found himself spending more time customizing databases to do what he needed than working on his original application.
“We spend so much time building accidental databases, we said why not build one on purpose?” Horowitz said during an interview on the opening day of MongoDB World in New York City yesterday.
And today he took to the stage for a keynote speech in which he unveiled new features of MongoDB, the hugely popular NoSQL database.
He couched the announcement by saying the improvements—in three main areas—might not seem to be huge technological breakthroughs, but they do provide a means for developer productivity and operational correctness.
First, on the server-side, Horowitz talked about document-level locking, which enables greatly improved concurrency. This will be added in the 2.8 release of the database, which he said should be out in “late fall.” This more granular locking will allow for more simultaneous writes to occur without locking up resources or hurting performance, he explained.
Also due out in the 2.8 release are new storage engines, each with different performance characteristics that can be selected based on use case, Horowitz said. These dedicated storage engines, for things such as encryption or compression, can be run simultaneously in single replica sets, he said.
The last improvement he discussed is in automation, as new managements tools for orchestrating large MongoDB clusters arrive that enable rolling configuration changes out of the box. Provisioning today is only available for Amazon Web Services, but other platforms are in the works.