Which Engine?

A quick rundown of which GCP computing option makes the most sense to use for different use cases.

Transcript

1. @tpryan
   Which Engine?
   Terrence Ryan
   @tpryan
   

   View Slide

2. @tpryan
   Introduction
   

   View Slide

3. @tpryan
   Who are you?
   Who are you?
   

   View Slide

4. @tpryan
   Computing
   Virtual machines Containers Serverless
   

   View Slide

5. @tpryan
   Virtual Machines
   

   View Slide

6. @tpryan
   Why Virtual
   Machines?
   ● Recreate existing environments
   ● Leverage existing skill sets
   ● Run specialized software
   

   View Slide

7. @tpryan
   Virtual machines
   Compute Engine
   ● 1 - 64 processors
   ● .6 - 416 GB (.9 - 6.5 GB per proc)
   ● Disk limits
   ○ 65 TB Persistent
   ○ 3 TB SSD
   ○ 208 RAM
   ● Preconfigured images
   ○ Can build your own
   ● Spin up in 10s of seconds
   

   View Slide

8. @tpryan
   Compute
   Engine demo
   Kick the tires and launch a
   few VMs
   

   View Slide

9. @tpryan
   Virtual machines
   Cloud Launcher
   ● Prebuilt software images
   ● Single systems
   ● Groups of systems
   

   View Slide

10. @tpryan
    Containers
    

    View Slide

11. @tpryan
    Why Containers?
    ● Performance
    ● Repeatability
    ● Portability
    

    View Slide

12. @tpryan
    Containers
    Container Engine
    ● Managed Kubernetes
    

    View Slide

13. @tpryan
    Kuber-what-es?
    Kuber-what-es?
    

    View Slide

14. @tpryan
    Frontend
    10.0.0.1
    Public
    130.91.x.x1
    API
    10.0.0.2
    Public
    130.91.x.x2
    Data
    10.0.0.3
    Kubernetes
    ● Container Orchestration System
    ● Allows for running a large number of containers in
    production
    ● Organizing them into workloads
    ○ Microservices
    ○ Scheduled jobs
    ○ Batch jobs
    

    View Slide

15. @tpryan
    Kubernetes demo
    Whack a mole
    

    View Slide

16. @tpryan
    Kubernetes
    DNS
    Server
    Logging
    Service
    Discovery
    Monitoring
    !
    

    View Slide

17. @tpryan
    DNS
    Server
    Logging
    Service
    Discovery
    Monitoring
    !
    Kubernetes
    

    View Slide

18. @tpryan
    Containers
    Container Engine
    ● Managed Kubernetes
    (Containers as a service)
    ● Smart defaults for
    ○ Monitoring
    ○ Logging
    ○ DNS
    ○ Service Discovery
    ● Auto scaling
    App Engine flexible environment
    ● Custom Runtime
    ○ Bring your own Dockerfile
    ● Autoscaling
    ● Given one endpoint per service
    

    View Slide

19. @tpryan
    Containers
    Container Registry
    ● Container image storage
    ● Works with
    ○ GKE
    ○ App Engine flexible
    ○ Docker
    Container Builder
    ● Container building as service
    ● Integrates with Container Registry
    ● Integrated with build processes
    

    View Slide

20. @tpryan
    Serverless
    

    View Slide

21. @tpryan
    Why Serverless?
    ● Systems Operations != Development
    ● Metered costs
    ● Eliminates wrote plumbing
    

    View Slide

22. @tpryan
    Serverless
    App Engine flexible runtime
    ● Languages
    ○ Java 8 / Servlet
    3.1 /
    Jetty 9
    ○ Python 2.7 & 3.5
    ○ Node.js
    ○ Ruby
    ○ PHP
    ○ Go
    App Engine standard runtime
    ● Languages
    ○ Python 2.7
    ○ Java 7
    ○ PHP 5.5
    ○ Go
    ● Constrained capabilities
    ● Incredibly rapid scale
    

    View Slide

23. @tpryan
    App Engine
    Load demo
    Quick demo showing
    App Engine in comparison
    to Compute Engine
    

    View Slide

24. @tpryan
    Serverless
    App Engine flexible runtime
    ● Languages
    ○ Java 8 / Servlet
    3.1 /
    Jetty 9
    ○ Python 2.7 & 3.5
    ○ Node.js
    ○ Ruby
    ○ PHP
    ○ Go
    App Engine standard runtime
    ● Languages
    ○ Python 2.7
    ○ Java 7
    ○ PHP 5.5
    ○ Go
    ● Constrained capabilities
    ● Incredibly rapid scale
    Cloud functions
    ● Javascript / Node.js
    ● Respond to events
    

    View Slide

25. @tpryan
    Cloud Functions
    Demo
    Very Quick Example of Cloud
    Functions
    

    View Slide

26. @tpryan
    Conclusions
    

    View Slide

27. @tpryan
    Virtual machines Containers Serverless
    Portability at the cost of
    efficiency
    Efficiency at the cost of
    portability
    Current industry compromise
    

    View Slide

28. @tpryan
    

    View Slide

29. @tpryan
    

    View Slide

30. @tpryan
    Thank you
    terrenceryan.com
    @tpryan
    This preso: http://bit.ly/tpryan-compute
    

    View Slide