Qumulo Promises Insights into Data with New Software-Based Storage
Qumulo, the Seattle data storage startup that has raised $67 million, says it has about 15 customers so far for the high-capacity system it’s bringing to the market today.
For the last three years, the company founded by early employees of Seattle’s last major data storage success, Isilon, has been quietly at work on Qumulo Core—a software-based storage system designed to handle billions of digital files and objects, and, most importantly, help companies understand the data they are storing.
Co-founder and CEO Peter Godman says the system harnesses several trends and technology developments in data storage. First and foremost, he says, Qumulo designed its products in concert with customers who are concerned less about the storage itself and more about the information contained within it.
“For the people that administer data storage, the data is a bigger problem than the storage,” he says.
That’s a change from the 1990s, when the focus was on storage systems that were simply big enough to handle the masses of data companies were accumulating, often spread among many smaller containers. In the 2000s, storage providers such as Isilon built big buckets that could scale up to accommodate growing data hordes.
Today, with hundreds of users and machines dropping files in and pulling them out of the same big storage buckets, companies are struggling to understand things like what data they have, how it’s being used and by whom, and why their storage capacity can be gobbled up in the course of a weekend.
“Nothing is better-positioned than storage to say something useful about data,” Godman says.
Godman says the Qumulo Scalable File System, for example, automatically creates a database that allows real-time analytics of the data stored within it. It can show things like storage capacity and performance—the speed at which files go in and out of the storage—through the lens of the data. A view from the file system’s dashboard instantly shows which directories are consuming the most storage capacity, or whether a certain workflow is sending too many requests to the storage system and gumming up the works for everything else trying to access it.
Qumulo can do this because it focused on flash storage, the solid-state storage medium in smartphones and tablets, as opposed to spinning disk drives—though it includes those, too. Flash offers superior performance and allows quick access to the metadata on which the database is built, Godman says.
“The flash lets you build a database into the product itself,” Godman says. “You never really could have done that before because of the limitations of spinning disk.”
Another change is Qumulo’s focus on software first, while the norm in traditional data storage was selling proprietary storage hardware. While Qumulo is marketing its own branded storage servers, the software can run on any commodity hardware, hardware from other vendors, virtual machines, and cloud-based machines, Godman says. So far, all but one of Qumulo’s customers have chosen its hardware, he adds.
A base system costs $50,000, which includes four of Qumulo’s storage servers and a year license to the software to run it. It can store 100 terabytes of data and can scale up to petabytes of storage easily, Godman says.
Qumulo, which announced a $40 million funding round led by Kleiner Perkins Caufield & Byers earlier this year, says its customers include major animation studios Ant Farm and Blind, fossil fuels giant Sinclair Oil, and medical data efforts such as the University of Washington’s Institute for Health Metrics and Evaluation.
Qumulo brings its product to market among a crowded field of data storage startups including Igneous Systems, Pure Storage, DataGravity, and Tamr. CB Insights says startups in this category raised $1.62 billion last year.