DBOS

Old operating systems run the world. Well into their fourth decade of use, Linux and Windows now require a crazy quilt of patches, logs and managers to keep them operating at the colossal scales of today’s cloud environments. Performance, reliability and security all suffer with the resulting complexity. To compensate, enterprise developers bolt on dozens of subscription tools and employ large ops teams—only to further expose their systems to debilitating cyberattacks, with potential recovery times of weeks or even months. “We’ve reached a critical turning point, where the more we try and protect these old operating systems, the more surface area and risk we’re creating,” says Andy Palmer, co-founder and Chairman of DBOS. “It is time for an evolution in operating system architecture.”

DBOS, a startup rooted in world-leading academic research conducted at Stanford, MIT, and Carnegie Mellon, is using the modern database as the foundation of an entirely new operating system. As computing applications have grown to incorporate thousands of machines across broad geographies, DBOS solves the grand challenge of maintaining ”state”—the system’s awareness of its own status and modifications—by making the database its fundamental layer. It embraces the popularity of microservices architectures, and delivers infrastructure that is elastic and serverless, but also stateful. “At DBOS, we are taking the most significant innovations in systems engineering over the last thirty years, and incorporating them into the most modern operating system we can imagine,” says Palmer.

DBOS—or “database operating system”—flips the fundamental underlying abstraction: rather than assuming ”everything is a file,” thereby requiring ever-expanding logs and management tools to keep up with today’s multi-cloud scales, DBOS uses “everything is a table,” making state its foundational layer. “DBOS runs the operating system on top of the database—rather than the current practice, which is to run the database system on top of the operating system,” explains co-founder Mike Stonebraker, MIT computer science professor and winner of the 2014 A.M. Turing Award.

DBOS achieves security through resilience. For companies with huge amounts of valuable data—and a consummate threat of attack—DBOS offers a fundamentally new level of infrastructure stability. While existing solutions are primarily focused on detection, DBOS acknowledges that evolving AIs will generate more and novel attacks, far beyond what human-authored detection tools can defend. Its clean-sheet design, offered as a Function-as-a-Service, minimizes the surface area of the system, and therefore its vulnerabilities. But more importantly, DBOS’ unique approach to system state allows for nearly instantaneous resilience. “With DBOS, you can quickly roll back your applications to any point in time,” says co-founder Qian Li. By enabling that quicker reaction, DBOS minimizes the amount of damage intruders can do. And by avoiding the need to rebuild a compromised system, DBOS offers resilience that is essentially seamless, and an efficient method to deal with a neverending list of cyber threats.

Over the last decade, banks, insurance companies, and information-driven government agencies have increasingly struggled to keep their rapidly expanding systems stable and secure—within the entanglements of convoluted legacy code. ”With DBOS, we have gone back to the foundational layer of the operating system to create a fundamentally more resilient infrastructure,” says Palmer.