GopherTok

English | 简体中文

	The sixth Bytedance Youth training camp big project works, a simple version of Tiktok project ，built with go-zero microservice . Completed by the gopher team

Architecture

🚀Technology stack

Feature	Implementation
HTTP framework	go-zero
RPC framework	go-zero
ORM framework	gorm
Database	Innodb-cluster,redis-cluster,mongodb
Object storage	Tencent Cloud COS, Minio
Service discovery, registration, and configuration center	etcd,nacos
Tracing	jaeger
Service monitoring	prometheus,grafana
Message queue	kafka
Log collection	filebeat,go-stash,elasticsearch,kibana
Gateway	traefik
Deployment	Docker,docker-compose
CI/CD	Github Action

High availability

• innodb-cluster

• redis-cluster

• minio cluster

• 4 nodes

• kafka cluster

• 3rd node

• The APIs and RPCs of the services start multiple docker instances. Traefik is used to load balance the APIs, and the RPCs implement load balancing via etcd to ensure service reliability and easy scalability during peak periods.

High concurrency

The RPC calls of each service are executed concurrently, significantly reducing response time. Redis is used as a cache for high-frequency data, reducing the pressure on MySQL. Kafka is used to asynchronously write to MySQL, increasing system throughput.

High performance

kafka uses clustered writes to greatly reduce disk io and network io

Configuration center and service discovery/registry center

Nacos is used as the configuration center

Etcd is used for service discovery and registry

Tracing

Jaeger is used for link tracing across services.

Log collection

Filebeat collects logs and sends them to Kafka for buffering. Go-stash filters the logs based on configuration and outputs the filtered logs to Elasticsearch. Kibana is responsible for visualizing the logs.

Supervisory control

Prometheus is used for service monitoring, with Grafana providing a visualization interface.

visualized by grafana

Gateway

Traefik is used as the gateway, load balancing requests to the API containers of each service based on routing rules.

load-balanced to various service 'api' container instances

CI/CD

• Use GitHub Action for CI/CD and automate testing after each commit Then you can manually build the image of each service, and automatically push it to the dockerhub after it is built Then ssh into the remote server, using the new image and already written 'docker-compose' automatically deploy the new container

Thanks

字节跳动青训营

Licence

GopherTok is open-source under the MIT license. Please follow the project for updates.