Google’s cluster management system that preceded Kubernetes was Borg. Borg was Google’s internal system for scheduling and managing large numbers of applications across clusters of machines.
Borg handled tasks such as resource allocation, workload scheduling, monitoring, service discovery, and restarting failed processes. It allowed Google to run diverse services efficiently at enormous scale, long before Kubernetes became widely available to the public.
Kubernetes was announced by Google in 2014 and was strongly influenced by Borg’s design and operating experience. Several Kubernetes contributors had previously worked on Borg, although Kubernetes was redesigned as a more broadly usable open-source platform and was written in Go rather than Borg’s C++.
A common mix-up is Apache Mesos, another important cluster manager. Mesos was developed independently and was not Google’s predecessor to Kubernetes. Google also had a later internal system called Omega, which explored a different scheduling architecture, but Borg is the canonical answer to this question.