cloud-native-go

README

cloud-native-go

Scratch repository for implementing cloud native Go patterns

Running the gRPC Server: make serve

Building CLI client to communicate with the gRPC Server: make build-client

Run a Hello World example: Ensure gRPC server is running and CLI binary is built make send

Generate protofile: make generate

Stability Patterns

Circuit Breaker

To demonstrate the CircuitBreaker pattern, start the gRPC server with make serve. In another terminal, run make circuitbreaker. If using the CLI tool, you can run bin/client cb and pass custom parameters to tweak the demo output.

Observe the log output from the server which showcases the Circuit Breaker in action!

Debounce

To demonstrate a Debounce pattern, start the gRPC server with make serve. In another terminal, run make debounce. If using the CLI tool, you can run bin/client debounce and pass custom parameters to tweak the demo output.

Observe the log output from the server which showcases restricting clusters of requests.

Retry

To demonstrate a Retry pattern, start the gRPC server with make serve. In another terminal, run make retry. If using the CLI tool, you can run bin/client retry and pass custom parameters to tweak the demo output.

Observe the log output from the server which showcases retrying given transient errors.

Throttle

To demonstrate a Throttle pattern, start the gRPC server with make serve. In another terminal, run make throttle. If using the CLI tool, you can run bin/client throttle and pass custom parameters to tweak the demo output.

Observe the log output from the server which showcases throttling excessive requests.

Timeout

To demonstrate a Timeout pattern, start the gRPC server with make serve. In another terminal, run make timeout. If using the CLI tool, you can run bin/client timeout and pass custom parameters to tweak the demo output.

Observe the log output from the server and the returned responses to observe that we can successfully wrap a timeout around a given slow function, even if that function is out of our control and doesn't accept a context.

Concurrency Patterns

Fan-In

To demonstrate a Fan-In pattern in which multiple input source channels can be multiplexed to a single destination channel, start with make serve. In another terminal, run make fanin. If using the CLI tool, you can run bin/client fanin and pass custom parameters to tweak the demo output.

Observe the log output from the server and the returned responses to see that all values can be concurrently read from the destination channel. The number of sources is customizable and each source will be randomly assigned a number of values to add to the channel from 1 to 10. You can visually map values that get added to the source channel and to the destination channel as indicated by the first digit.

In the example below, the value of 12 that is read of the destination channel indicates its from the 1st source channel by first digit. Similarily, the value of 22 that is read from the destination channel indicates its from the 2nd source channel:

2021/11/21 13:17:56 ➕ Adding 12 to source channel #1 ➕
2021/11/21 13:17:56 📖 Reading 12 off destination channel 📖
2021/11/21 13:17:56 📖 Reading 22 off destination channel 📖
2021/11/21 13:17:57 ➕ Adding 13 to source channel #1 ➕

Fan-Out

To demonstrate a Fan-Out pattern in which a source channel can split its values between multiple competing destination channels, start with make serve. In another terminal, run make fanout. If using the CLI tool, you can run bin/client fanout and pass custom parameters to tweak the demo output.

Observe the log output from the server to see that different channels are getting distinct source values. It noteably isn't implemented as round-robin so some destination channels may receieve multiple and some may receive none. They compete to avoid the issue that could occur if a destination channel isn't ready to receive yet (a slow worker).