Py_gRPC

gRPC implementation in Python.

Locust Tool - Load Testing Report

• gRPC load testing report:
• REST load testing report:

Requirements

• Python 3.6.15

Project Setup and Execution

• clone py_grpc project
• cd py_grpc
• python -m venv venv
• source venv/bin/activate
• pip install -r requirements.txt

gRPC

• In REST call --> request_message --> HTTP payload --> TCP payload --> etc.
• But In gRPC call --> request_message --> TCP payload --> etc.
• So when a grpc_request_message reaches server (TCP payload), the expectation is GRPC service should directly get this TCP payload and proceed.
• But now what's happening is grpc_request_message reaches server (TCP payload). Server passes it to uwsgi which thinks that it's a http protocol payload and tries to unpack it. This might be the reason for failure.

Unary gRPC

This is a simple gRPC which works like a normal function call. It sends a single request declared in the .proto file to the server and gets back a single response from the server.

Unary gRPC Setup and Execution

• [OPTIONAL] python -m grpc_tools.protoc --proto_path=./protobufs ./protobufs/unary.proto --python_out=./unary_grpc --grpc_python_out=./unary_grpc
• In one terminal --> python ./unary_grpc/unary_server.py
• In another terminal --> python ./unary_grpc/unary_client.py

Bidirectional gRPC

Both gRPC client and the gRPC server use a read-write stream to send a message sequence. Both operate independently, so gRPC clients and gRPC servers can write and read in any order they like, i.e. the server can read a message then write a message alternatively, wait to receive all messages then write its responses, or perform reads and writes in any other combination.

NOTE: gRPC guarantees the ordering of messages within an individual RPC call. In the case of Bidirectional streaming, the order of messages is preserved in each stream.

Bidirectional gRPC Setup and Execution

• [OPTIONAL] python -m grpc_tools.protoc --proto_path=./protobufs ./protobufs/bidirectional.proto --python_out=./bidirectional_grpc --grpc_python_out=./bidirectional_grpc
• In one terminal --> python ./bidirectional_grpc/bidirectional_server.py
• In another terminal --> python ./bidirectional_grpc/bidirectional_client.py

REST and gRPC Dual Mode

This POC shows working simultaneously with both REST and gRPC services listening at different ports in the same application.

Advantages of REST and gRPC Dual Mode POC

• Decouples core business logic and data communication logic (SOAP, REST, gRPC, etc.).
• Application can be extended to any type of data communication format with minimal changes without affecting business logic.
• Generalizing application structure (standardization) which brings many possibilities and helps in many scenarios like...
  • Supporting many data communication formats at a time.
  • Migrating from one data communication format to another with minimal effort, taking one chunk at a time. No need to convert all services to another format in one go. Which helps in backward compatibility for older data communication format without affecting the application functionalities.

Load Testing using Locust Tool

Locust tool is used to perform load testing on the application via both REST and gRPC communication.
Based on the Locust performance reports, I found that performance of gRPC communication is much better than the performance of REST communication.

Reports Path: rest_grpc_dual_mode/tests/load_tests/reports/

Locust Commands

• [OPTIONAL] python -m grpc_tools.protoc --proto_path=./protobufs ./protobufs/rest_grpc_dual_mode.proto --python_out=./rest_grpc_dual_mode/tests/load_tests/grpc/generated --grpc_python_out=./rest_grpc_dual_mode/tests/load_tests/grpc/generated
• [For REST] locust -f rest_grpc_dual_mode/tests/load_tests/rest/rest_load_test.py --host http://localhost:5000 --users <MAX_USERS> --spawn-rate <NO_OF_USERS_TO_ADD_PER_SEC>
• [For gRPC] locust -f rest_grpc_dual_mode/tests/load_tests/grpc/grpc_load_test.py --host http://localhost:50051 --users <MAX_USERS> --spawn-rate <NO_OF_USERS_TO_ADD_PER_SEC>

References

• Implementing gRPC In Python
• gRPC documentation
• Python gRPC documentation
• Python gRPC API reference
• Protocol Buffers
• gRPC Motivation and Design Principles
• BloomRPC tool - GUI Client for GRPC Services
• Werkzeug - Werkzeug is a comprehensive WSGI web application library
• Locust Tool
• Locust Documentation
• Locust - Testing non-HTTP systems (gRPC)
• gRPC benchmarking and load testing tool