It used to be the case that we couldn’t find anything on the internet. We clicked around link to link searching for what we wanted. Yahoo! saw this problem and first settled on the Dewey Decimal system as a model to solve this problem, using humans to create an exhaustive table of contents for the internet. Google looked at the problem and saw a machine-generated and curated index of everything. Google won. The more a startup knows about technology, the better chance it has of creating a killer product. Technology is the mechanical advantage of humans… it’s what lets a small group of humans improve the condition of a large group. Startups without technology may be greeted warmly by a few customers, but they can’t change the world.
But when a startup only understands technology, it inevitably invents things with no compelling use. In my last startup, we observed that we could control nondeterminism in the execution of a CPU. There had to be a use for that! In hindsight it’s not clear that there was. Over time, as it became apparent that we hadn’t done our market homework, we went through the miserable, grinding process of realizing that we would never take off. Eventually (probably later than we should) we capitulated and sold. The team that we loved working with, and the technology that we poured our lives into, was gone. Startups that understand technology, but not necessity, die a slow painful death.
The world is bettered by useful invention. Useful invention is the child of need and technology.
When we built Qumulo, we didn’t start off with a vision for a product. We started off with a vision of how to build a product that people cared about.
First, we built a machine to understand necessity. Our product team conducted user groups 18 months before we had a product. We hosted many hundreds of research calls with end users of data storage. We conducted more than a thousand interviews on the minutiae of what people want out of data storage. We road-tested ideas and brought data storage users into our environment regularly. We spent time with users, looking over their shoulders, buying them coffee, and trying to live their lives with them, every step of the way. Eventually, we started to learn the necessities of being a storage administrator.
Second, we built a team that understands technology. We hired some of the best software engineers in the industry, and had them work side by side with product owners every day, and asked them to continually map the space of necessity against the space of technology, juggling and ideating.
Third, we built a culture that allows us to rapidly try out and experiment with combinations of necessity and technology. We’ve learned how to ship robust enterprise infrastructure software every two weeks, and gather feedback on our software continually and iteratively. From this, we learned:
Storage matters, but data matters more. The storage industry creates waste by making storage opaque. The unsung hero of data reduction is not storing data. Storage administration is giving way to data administration.
High performance storage doesn’t need custom hardware. The choice between easy scale and small file and transactional performance is a false one. We can build and deliver robust infrastructure in two-week increments.
From these realizations we’ve built Qumulo Core, the world’s first Data-Aware Scale-Out NAS. Today we’re announcing what we’ve built, who’s using it, and what they’re using it for. We think we’ve married necessity and technology in a way that will upend an entire industry.
We can’t wait for you to see what we’ve built.
Peter Godman uses his expertise in distributed file systems and high-performance distributed systems to guide product development and management at Qumulo. He is the inventor of more than 20 granted patents in the areas of file system design, distributed systems, and shared-memory concurrency. As Qumulo’s founding CEO, Peter led the company through fundraising rounds totaling $100M, the delivery of Qumulo Core, and the acquisition of Qumulo’s first hundred customers. Prior to Qumulo, Peter served as CEO of Corensic, which brought the world’s first thin hypervisor to market. Before that, he was Director of Software Engineering at Isilon, where he led its file system team through several major releases.