Deployment Automation with Microservices
- 1. SPA (browser) Mobile Apps App Services DB Server Integration Server Auth Server Document Management Server Lets start out with a conventional architecture
- 2. SPA (browser) Mobile Apps App Services DB Server Master Integration Server Auth Server DMS Real world deployment and clustering for failover and scalability brings in complexities App Services Load Balancer DB Server RO Slave DB Server RO Slave Integration Server DMS Load Balancer Load BalancerLoad Balancer
- 3. SPA (browser) Mobile Apps App Services DB Server Master Integration Server Auth Server DMS Microservices Architectures further distribute processing and make the architecture complex Service1 Load Balancer DB Server RO Slave DB Server RO Slave Integration Server DMS Load Balancer Load BalancerLoad Balancer App Services Service2 Load Balancer App Services Service3 Load Balancer DMSNOSQL Server Load Balancer App Services Service4 Load Balancer DMSApache Spark + HDFS Load Balancer App Services Service5 Load Balancer API Gateway Inter Service Communication Bus
- 4. Managing/Automating tens of VMs, through conventional scripting is tough, error prone and unreliable If your deployment automation scripts are aware of nature of application inside the VM, then it makes your deployments brittle and not easily scalable Besides, other softwares are needed which keep track of which machines are running which VMs and for automating dynamic scale-out and failover Besides a good deployment automation also needs to support, service discovery, scheduling(which VMs go where), healthchecks and monitoring, upgrades and rollback support and easy scale out of selective VMs Benefits of containerization The container is quite opaque to the deployment orchestration scripts DEV, QA, UAT, PRE-PROD and PROD, all environments can be treated alike and this facilitates CI and CD Container orchestrators like Kubernetes and Mesos can satisfy all the deployment automation requirements like service discovery, scheduling(which VMs go where), healthchecks and monitoring, upgrades and rollback support and easy scale out of selective VMs