Skip to content

Boavizta/ecobenchmark-applicationweb-backend

Repository files navigation

This repository is in stand by, the project continues with another goal : provide data to Boavizta / efootprint builders

==> https://github.com/Boavizta/ecobenchmark-4-builder-efoootprint

Eco benchmark

This repository is benchmarking different scenario to try to compare the energy consumption, for several languages.

The different scenario will be the following, for each language:

  • default, optimised but not extremely
  • different degraded use cases.

Eco Benchmark diagram

Workflow

The complete workflow is based on docker images. Every image should be named as <org>/service-<service_name>:<use_case>.

Each image can be built using the following command.

# oRG=jdrouet is used by default but can be changed by exporting a different variable.
./builder/service.sh <name_of_your_service> <use_case>

Starting the database

Postgres

to run the migrations when starting the database

--volume $(pwd)/migrations:/docker-entrypoint-initdb:ro \

today, the 30th of april 2022, this is the latest release

postgres:14.2-bullseye

docker run -d \
  --name eco-benchmark-database \
  # to run the migrations when starting the database
  --volume $(pwd)/migrations:/docker-entrypoint-initdb.d:ro \
  --port 5432:5432 \
  -e POSTGRES_PASSWORD=mysecretpassword \
  postgres:14.2-bullseye

export DATABASE_URL=postgresql://postgres:mysecretpassword@127.0.0.1:5432/postgres

Use Case Status

Langage Reference (main) No Index ORM Loop MySQL GRPC No Pagination Aggregation on code side
Go Lang X X X X X X X
Rust X X X X X
PHP (Symfony) X X X X N/A X X
JVM/Kotlin X X X X X X X
Node.js X X X X X X
Ruby On Rails
JVM/Java (Quarkus) X
JVM/Java (Quarkus-Reactive) X
Native/Java (Quarkus) X

Use Case Documentation

Default

Branch : main

Description : classic optimized implementation, but not too much.

No Index

Branch : usecase-no-index

Description : oups, we forgot to put the indexes. For thoses who are laughing, you all have people running wordpress website around you. Check the db...

ORM Loop

Branch : usecase-orm-loop

Description : for the API path GET /api/accounts/:account_id/lists/ we do a classic loop like developer usually do using a ORM or layered architecture.

Instead of one request, the get list become :

  • Get the list from the db (paginated)
  • Loop on the list
    • get the tasks of the lists
    • fill the tasks in the list object
  • return full list with task.

MySQL

Branch : usecase-mysql

Description : replace postgresql by mysql

Note : DATABASE_URL should be like : mysql://root:mysqlpw@tcp(localhost:3306)/ecobenchmark?parseTime=true

?parseTime=true is mandatory.

GRPC

Branch : usecase-grpc

Description : use grpc instead of classic http/json.

Based on this service definition : https://gitlab.com/jeremie.drouet/eco-benchmark/-/blob/usecase-grpc/service/go-pgx/endpoints/endpoints.proto

No Pagination

Branch : usecase-no-pagination

Description : for the API path GET /api/accounts/:account_id/lists/ remove pagination.

Aggregation on code side

Branch : usecase-aggregation-code-side

Description : for the API path GET /api/stats remove group by and aggregate on code side.

Contributing

Everybody is more than welcome to contribute to this benchmark! Please check out the Contributing guide for guidelines about how to proceed.

You can also join us and become a member of the non-profit French association.

License

The Eco benchmark is released under the AGPL-3.0 license.