GRPC - From Tutorial to Production

Explore the evolution of gRPC from tutorial to production in this 44-minute conference talk from GopherCon 2017. Dive into the best-in-class toolkit for inter-service communication, starting with a crash course on the Go tutorial code. Learn how to address advanced concerns when deploying gRPC in production environments, including monitoring, load balancing, error handling, and timeouts. Gain valuable insights from two years of production experience running ngrok's infrastructure with gRPC. Cover topics such as secure transport, protobuf serialization, client-server interactions, multitenancy, performance optimization, logging, interceptors, timeouts, retry logic, structured errors, defensive measures, streaming, and cross-language compatibility. Discover the challenges and future prospects of gRPC implementation in real-world scenarios.

Intro
HOW DO MICROSERVICES TALK TO EACH OTHER?
A HIGH-PERFORMANCE. OPEN-SOURCE UNIVERSAL RPC FRAMEWORK
LET'S BUILD A CACHING SERVICE
IMPLEMENT THE METHODS
RETURNING ERRORS
SECURE TRANSPORT
protobuf serialization (pluggable) Clients open one long-lived connection to a grpc server A new HTTP/2 stream for each RPC call
MISBEHAVING CLIENTS: MULTITENANCY
IT'S TOO SLOW
ADD LOGGING
CLIENT INTERCEPTOR
SERVER INTERCEPTOR
ADD A TIMEOUT
NEW REQUIREMENT: DRY RUN
NETWORKS FAIL
ADD RETRY LOGIC
NON-IDEMPOTENT OPERATIONS
FAILED OPERATIONS ARE TIMING OUT
IT'S NOT THE SERVER
STRUCTURED ERRORS
FEATURE REQUEST: DUMP
ADD DEFENSIVE MEASURES
NOT ENOUGH MEMORY
START STREAMING
IT WORKS ...
SCALE OUT WITH LOAD BALANCING
CALL IT FROM OTHER LANGUAGES
SO WHAT'S THE BAD NEWS?
THE FUTURE