TALK
THU MAY 7 • 1:45 PM - 2:30 PM
PERSENNE (MEZZANINE LEVEL)From Scale to Stability: High Availability in Temporal
This talk outlines how to design highly available systems with Temporal by treating failure as a constant and building resilience into every layer of the architecture. It covers core principles like eliminating single points of failure, leveraging replication and write-ahead logs, and ensuring reliability across proxies, cells, and dependencies. The session emphasizes that true stability comes from automation, observability, and continuous operational rigor rather than one-time fixes.
Liang brings over a decade of specialized expertise in distributed systems, helping teams build more reliable and scalable applications. At Temporal, he leads the team responsible for designing and developing Temporal's core replication stack, enabling extraordinary levels of availability beyond the capabilities of a single cluster. Through this replication stack, Liang's team delivers the global namespaces offering on Temporal Cloud, ensuring 99.99% availability and resilience against regional outages. Additionally, they develop internal tools that seamlessly migrate customer workloads between clusters. Before joining Temporal, Liang managed the Uber Cadence team and was an active contributor to the Cadence community.
Sergey Bykov is responsible for the architecture of Temporal Cloud. Before joining Temporal, Sergey was one of the founders of the Orleans project at Microsoft Research and led its development for over a decade. The mediocre state of developer tools for cloud services and distributed systems continues to fuel his passion for qualitatively improving developer productivity in this space.