Managed Kubernetes in Azure door Arjen Steinhauer

devNetNoord #13

Transcript

1. in Azure Managed Kubernetes
2. Even voorstellen… • Arjen Steinhauer asteinhauer@ilionx.com • Principle Consultant / Architect bij ilionx (Utrecht) • Werkzaam in IT sinds 1995 • Microsoft platform (.NET) • Microsoft Azure • Integration • Continuous Delivery / DevOps
4. 1. Maintaining existing applications 2. Re-deploying new versions of existing applications 3. Troubleshooting issues from feature additions to existing applications 4. Deploying a new application
5. 1. Maintaining existing applications 2. Fixing* existing applications 3. Adding features to existing applications 4. Fixing the features someone else added to existing applications 5. Building a new application (maybe … if there’s time!)
6. Set of IT Problems forever Patches break installations Deployment binders Lead to shadow IT, tension between dev/ops Etc Works on my machine
8. Moore’s Law Limit of vertical scale? (scale-up)
9. 256 core processor
10. Distributed Systems
11. Common approach • Client Server model • 3-tier architecture • Object Oriented (C#, java, C++, …) • RPC, DCOM, SOAP, WCF, REST, … Client Service DB Client Client
12. Service Service Service Service Service Middleware scale-out Client Service DB Client Client
13. Service Service Middleware scale-out Client Service DB Client Client Service Service Service Load Balancer
14. Service Service Database partitioning Client Service DB Client Client Service Service Service Load Balancer DB Cache
15. The problem of scale-out • Concurrency • Partitioning • High throughput Speed up number of nodes
16. “Speed-up” of traditional approach… • Reality is non linear • More nodes does not mean more scale 10 5 Speed up number of nodes
17. Need more scaling options  Scale Cube (3D) • X • Horizontal scale • More nodes • Network Load Balancing • Y • Functional scale • Micro services • Z • Data partitioning • Separate tenants • Separate by region Y X Z
18. Micro Services: the goal • Build easier Distributed Systems • Automatic 3D scaling is possible • Less skilled dev’s can build more complex systems • Save development and operation costs • Run same code on-premises and in Cloud
19. Service Service Micro Services Client Service DB Client Client Service Service Service Load Balancer DB Cache Stateless middle-tier Stateless front-end state
20. Service Service Micro Services Client Service A data Client Client Service Service Service B Service Locator data data warehouse Stateful middle-tier Stateless front-end Analytics/ disaster recovery
21. Micro Services
22. Micro Services bij Netflix
23. Containers in Azure
25. Containers = Operating system virtualization Container Container Container Traditional virtual machines = hardware virtualization VM VM VM Windows Server containers No different from Linux containers Container Container Container Hyper-V containers Isolation plus performance Container Container Container
29. Demo Pull Run Build … a container
30. VM Host VM Host VM Host VM Host But things get complicated quickly
31. Very Quickly
35. Balance of control and responsibility depends on the category of the service MOVE-IN READY Use immediately with minimal configuration SOME ASSEMBLY REQUIRED Existing services are a starting point, with additional configuration for a custom fit BUILD FROM THE GROUND UP Building blocks, create your own solution or apps from scratch Responsibility On- Prem IaaS PaaS SaaS Applications Data Runtime Middleware O/S Virtualization Servers Storage Networking Microsoft Customer
36. Notes: (*1) (*2) (*3) App service and “container apps” provides for orchestration of a single container apps. In the current state, this is a PaaS offering that abstracts away these concepts. (*4) DC/OS has a good-sized OSS community behind it, however it pales in comparison with K8 ecosystem. (*5) Docker as a container format is very popular and has good community engagement, Docker EE gets some benefits from being associated with it. However, it does not have a strong ecosystem by itself. (*6) DC/OS refers to enterprise DC/OS. ACS provides open source DC/OS
37. Azure Container Instance Container Service Service Fabric Web Apps Batch Mesos DC/OS
45. Node Master CLI Node Node UI API
46. Master Master Master Node Node Node Node Node Node
48. Demo Aye captain! Kubernetes
49. CI/CD and DevOps workflow
51.  Stores container images
52. Docker Hub and Docker Store Docker Trusted Registry Azure Container Registry Docker Registry https://docs.docker.com/registry https://github.com/docker/distribution
57. ilionx Hondiuslaan 46 3528 AB Utrecht T (088) 05 90 500 E info@ilionx.com www.ilionx.com