Latency Myth-Busting: Why “<50 ms Global Median” Is Science-Fiction
Blog post from PubNub
Events in global networks, such as game updates or telehealth alerts, inherently face latency challenges due to the immutable laws of physics, with light taking approximately 74 milliseconds to travel halfway around the Earth in a vacuum. While some vendors claim sub-50 ms global latency, these assertions often neglect physical constraints or focus solely on localized regions. PubNub distinguishes itself by transparently measuring and publishing the latency of events within its infrastructure, utilizing synthetic traffic rather than selective demos, and employing auto-scaling Kubernetes-based systems to maintain low latency. The company emphasizes the importance of acknowledging "last-mile" latency, which is beyond their control, and focuses on optimizing the latency they can influence, providing detailed measurements and engineered solutions to improve user experience. PubNub's advertised <30 ms latency pertains specifically to the time events spend within their network infrastructure, highlighting their commitment to transparency and reliability in real-time applications.