What if you could store your data in the cloud and run complex queries and analytics on it where it resides, without moving it? That's the question high-performance cloud infrastructure and big data analytics specialist Joyent looks to answer with its new Joyent Manta Storage Service.
Manta is a next-generation cloud object store and data services platform designed to bring compute and analysis capabilities directly to customers' data in the cloud.
"The killer app for the personal computer was the spreadsheet," says Bryan Cantrill, vice president of Engineering at Joyent, who has spent the past three years immersed in the creation of Manta. "That was the killer app because it represented the convergence of data and compute. It allowed business folks to run that data on their desks, and they didn't need a time-sharing program to do that."
Cloud storage, he says, brought together two other pillars of technology—data and the network—and combined them. Now, Manta goes a step further, Cantrill says.
Is Manta the Spreadsheet of the Cloud?
"I believe Manta represents the spreadsheet of cloud computing," Cantrill says. "It opens up vistas of ad hoc analysis of unstructured data. What we're doing is converging compute with data with the network. We're converging all three of those into a single facility. This is the holy trinity of compute, network and data unified into a single offering."
By removing the need to manage infrastructure and move data, Cantrill says Manta gives enterprises the capability to process big data faster and more easily, all while keeping it secure and at prices on par with Amazon.
"You really can't bring a product to market that is not going to be competitive on price," he says. "We think there's more to cloud computing than just price, but it's simply a non-starter to overprice a product. You'll pay S3 prices for the storage, and you pay what are effectively EC2 or Joyent Infrastructure as a Service prices for any compute you spin up on your data. You may spin up for a second or two seconds or three seconds. If so, you just pay for one or two or three seconds of compute."
For instance, he says, maybe you want to run periodic validation checks on your backups. It's typically not done because it would mean dragging all your backups back to compute. But, he says, it becomes trivially easy to do such validation tests with Manta—you just have to spin up compute on the backups where they reside for a few microseconds.
Manta gives you the capability to execute compute tasks including log analysis, search index generation, financial analysis and other data-intensive tasks without moving data or setting up compute clusters and processing software. Cantrill explains that code is brought in parallel to physical servers in secure containers while data is automatically merged using the industry-standard MapReduce pattern.
Sign up for Computerworld eNewsletters.