Real-time streaming APIs: From data center to internet clients
When designing APIs such as the new GCP Firestore real-time database and Google Assistant, how did Google decide which trade-offs to make? Wenbo Zhu dives deep into the challenges faced while deploying a real-time streaming API designed for clients from data centers to the internet and details the trade-offs API developers need be aware of when designing such an API.
Talk Title | Real-time streaming APIs: From data center to internet clients |
Speakers | Wenbo Zhu (Google) |
Conference | O’Reilly Open Source Software Conference |
Conf Tag | Fueling innovative software |
Location | Portland, Oregon |
Date | July 15-18, 2019 |
URL | Talk Page |
Slides | Talk Slides |
Video | |
When designing APIs such as the new GCP Firestore real-time database and Google Assistant, Google often focuses on the semantics of the underlying service and tries to avoid introducing any messaging-layer semantics such as failure recovery and resumption support. Wenbo Zhu dives deep into the lessons learned deploying a public cloud API designed for data center clients that will be consumed by a variety of internet clients. It’s important to decide how to make those design trade-offs and avoid the need to support two different sets of APIs, which would be difficult to maintain. The solution often comes down to a rich client library to minimize the overhead or complexity, which would otherwise be forced to data center clients.