Microservices in practice with JHipster
From dev 💻 to prod 🚀
Pierre Besson
Paris JHipster Meetup #10, October 4th, 2017
Who am I ?
- Java developer @IpponTech
- FullStack & DevOps
- JHipster core team member
- Favorite technologies:
- ☕ JAVA EE
- 🍃 Spring
- 🐳 Docker
- ☸️ Kubernetes
- 300 collaborators worldwide (France, USA, Australia)
- 4 cities in France : Paris, Nantes, Lyon, Bordeaux
- Specialized in Java, Cloud, Big Data
The JHipster team
https://jhipster.github.io/team/
Introducting Microservices
Why microservices?
Microservices patterns
Beware microservice hype
Why microservices
- “Microservices are small, autonomous services that work together” (Sam Newman)
- Take the concepts of distributed systems and SOA to the next level
- Emphasis on :
- Technology Heterogeneity
- Evolutionary architecture
- Resilience
- Ease of deployment
- Organizational Alignment
- Composability
Microservices patterns
microservices.io/patterns/microservices.html
Beware microservice hype
Think twice before jumping on the microservice hype train
- Microservices are not simpler than monoliths
- Choosing the right decomposition is hard
- “Architect’s Dream, Developer’s Nightmare”
- Good idea to start with a monolith
- You are not Google/Netflix
JHipster as a microservice platform
JHipster goals
Extending JHipster for microservices
Monolith vs Microservices
Technologies used
The JHipster Microservice Stack
The JHipster Microservice Stack
JHipster goals
- 🏗️ Provide scaffolding for a Full Stack application
- 💠 Based on modern technologies
- 🔧 Maximize development productivity with tooling
- 📈 Production ready (maintainable, secure, scalable)
JHipster
from
monolith
to
microservices
Extending JHipster for a microservice architecture
- 🏗️ Generate several independent microservices and front-ends
- ️ ️Work out of the box in dev and in prod !
- Configure service discovery and centralized configuration
- 🔒Provide a distributed security solution
- Generate entities back-end and front-end separately
- Be DevOps ready: provide configuration for docker, CI-CD, ...
- Provide management dashboards
- Offer a Monitoring solution to manage production
Monolith vs Microservices
- Monolith:
- One application (Back-end + Front-end)
- Microservice architecture:
- Registry
- Microservices (back-end only)
- API Gateway (front-end + proxy)
- (Optional) OAuth2 server: UAA, Keycloak
Technologies used by JHipster Microservice
Rely on Spring Boot and Spring Cloud abstractions:
- Service discovery: Eureka/Consul
- Centralized configuration: Spring Cloud Config
- ⛩️ API Gateway: Zuul
- 🔀 Client side load balancing: Ribbon
- ⚡ Circuit Breaker: Hystrix
The JHipster Microservice Stack
Added value added over the base Spring Cloud stack:
- JHipster Registry: Eureka + Config server + Management
- ⛩️ JHipster Gateway: API management with Zuul and Swagger docs
- 🔁 CI-CD Configuration: Pipelines for Jenkins 🤵, Travis 👷, Gitlab
- 🚀 Deployments: Docker 🐋, Kubernetes ☸️, Rancher 🐄
- JHipster Console: logs + metrics + traces (based on ELK and Zipkin)
JHipster Microservice Architecture
jhipster.tech/microservices-architecture/
Show time
🎆Using JHipster to develop microservices
Running the full architecure with docker
Let's go to production
🚀Deploying microservices on Kubernetes
Introduction to microservice monitoring
Problems and solution
The 3 ways of observability
The JHipster Console
The problems
- Distributed system -> More complex to observe
- Logs are dispersed in many many log files
- Hard to locate the microservice that caused a problem
- Hard to follow the chain of requests
- 🐇/🐢 Studying latency is hard
- 💂 Resiliency mechanism are hard to configure properly
The solution
Comprehensive monitoring
- Logs : discrete event
- Metrics : numerical values (business or technical)
- Traces : chains of calls in the system (spans)
- Store everything in one place, with proper metadata
- Use a search engine to query and aggregate monitoring data
- Visualize using graphs and dashboards
- Navigate around
- Create alerts
The 3 ways of observability
The 3 ways of observability
The JHipster Console
www.jhipster.tech/monitoring/
- Setup the ZELK stack (Zipkin, Elasticsearch, Logstash, Kibana) in docker
- Enable reporting from any JHipster app with a few properties
- Logs + metrics forwarded with logback-logstash-encoder
- Traces are forwarded to Zipkin using Spring Cloud Sleuth
- Logs are enriched with trace information (Trace ID)
- Data is stored in Elasticseach for a certain number of days (Curator)
JHipster Console demo
Navigating around Kibana
Creating custom visualizations and dashboards
Graphing metrics, manipulating timeseries
Following call traces accross services (Kibana + Zipkin)
Questions ?
Slides available online
pierrebesson.github.io/pres-jhipster-meetup-10/