July 20th, 2016
(written by lawrence krubner, however indented passages are often quotes). You can contact lawrence at: firstname.lastname@example.org
Google has been using a single general-purpose RPC infrastructure called Stubby to connect the large number of microservices running within and across our data centers for over a decade. Our internal systems have long embraced the microservice architecture gaining popularity today. Having a uniform, cross-platform RPC infrastructure has allowed for the rollout of fleet-wide improvements in efficiency, security, reliability and behavioral analysis critical to supporting the incredible growth seen in that period.
Stubby has many great features – however, it’s not based on any standard and is too tightly coupled to our internal infrastructure to be considered suitable for public release. With the advent of SPDY, HTTP/2, and QUIC, many of these same features have appeared in public standards, together with other features that Stubby does not provide. It became clear that it was time to rework Stubby to take advantage of this standardization, and to extend its applicability to mobile, IoT, and Cloud use-cases.
Among the important goals:
Services not Objects, Messages not References – Promote the microservices design philosophy of coarse-grained message exchange between systems while avoiding the pitfalls of distributed objects and the fallacies of ignoring the network.
Payload Agnostic – Different services need to use different message types and encodings such as protocol buffers, JSON, XML, and Thrift; the protocol and implementations must allow for this. Similarly the need for payload compression varies by use-case and payload type: the protocol should allow for pluggable compression mechanisms.
Streaming – Storage systems rely on streaming and flow-control to express large data-sets. Other services, like voice-to-text or stock-tickers, rely on streaming to represent temporally related message sequences.