Edge Computing for WebApp

Explaining brief a history of edge computing for webapp

Transcript

1. Edge Computing for the
   WebApp
   ragingwind
   

   View Slide

2. Edge Computing
   

   View Slide

3. History of Computing
   

   View Slide

4. Mainframe
   

   computing
   - 1960 ~ 1970
   
   
   - Large, monolithic systems
   
   
   - Physical datacenter on-site
   
   
   - Serving via terminals
   

   View Slide

5. Server / Client
   

   computing
   - 1980 ~ 1990
   
   
   - Primary datacenter for large
   project and data storage
   
   
   - Working on the network for
   multiple client connections
   
   
   - Serving via local computer
   

   View Slide

6. Cloud
   

   computing is
   - 2000 ~ 2010
   
   
   - Data centers as public cloud platform not
   on-site
   
   
   - All of the processing and storage
   happening o
   ff
   site in the cloud
   
   
   - Running many of services, Infra, Services,
   Containers, Workers, Functions ...
   
   
   - AWS, GCP, Azure, ...
   

   View Slide

7. Edge Computing
   
   
   Today
   

   View Slide

8. Edge Computing
   - Ultra-low latency and decrease in bandwidth use
   
   
   - Decrease in server resources
   
   
   - Functionality not only static resources
   
   
   - Mobile, IoT, Apps need for speed
   

   View Slide

9. Edge Computing
   
   
   for the WebApp
   

   View Slide

10. Edge Computing for the
    WebApp
    https://vercel.com/blog/regional-execution-for-ultra-low-latency-rendering-at-the-edge
    

    View Slide

11. https://vercel.com/blog/regional-execution-for-ultra-low-latency-rendering-at-the-edge
    

    View Slide

12. https://www.prisma.io/blog/database-access-on-the-edge-8F0t1s1BqOJE
    Database access on the Edge
    

    View Slide

13. Kelsey Hightower
    'You can host not just images but logic and
    run it safely on the Edge, like why now ?'
    

    View Slide

14. Innovations for the Edge
    - Process Isolation
    
    
    - Optimized minimalistic code
    
    
    - HTTP default protocol
    

    View Slide

15. Process Isolation
    - Next process model after container, worker on cloud
    
    
    - Running in sandbox, isolated and stateless
    
    
    - Starting up really fast
    
    
    - Executing optimized minimalistic code
    
    
    - V8 Isolates(Vercel, Deno, Cloud
    f
    lare), microVMs(AWS
    f
    irecracker),
    WASI(WebAssembly) gVisor, OSv
    

    View Slide

16. https://aws.amazon.com/blogs/aws/
    f
    irecracker-lightweight-virtualization-for-serverless-computing/
    

    View Slide

17. https://developers.cloud
    f
    lare.com/workers/learning/how-workers-works/
    

    View Slide

18. https://deno.com/blog/anatomy-isolate-cloud
    

    View Slide

19. https://blog.jdriven.com/2022/08/wasi-capability-based-networking/
    

    View Slide

20. Optimized Minimalistic Code
    - Generate highly optimized machine code
    
    
    - low-level incremental computation engine for individual
    functions
    

    View Slide

21. https://turbo.build/pack
    

    View Slide

22. HTTP Default Protocol
    - `Web is becoming the computer then the concept of http being the
    most dominant protocol` - Kelsey Hightower
    
    
    - 0-RTT(Zero Round Trip Time) via TLS 1.3 or QUIC
    
    
    - HTTP/3 powered by `UDP`
    
    
    

    View Slide

23. HTTP/3 powered by `UDP`
    https://blog.cloud
    f
    lare.com/the-road-to-quic/
    

    View Slide

24. Edge Computing for WebApp
    - Using functions on the Edge as like local package
    
    
    - Using server resources on the cache of Edge as like local
    f
    iles
    
    
    - WebApp/Functions can be served in compact size as much as decreased
    
    
    - Bring computing power from outside Browsers in a second
    
    
    - Centralized WebApp can be running close to users
    
    
    - ...
    

    View Slide

25. Chronology of the WebApp
    

    View Slide

26. Image converting WebApp
    

    View Slide

27. WAS on Bare-Metal
    
    
    - Using computing power of
    dedicated physical server
    
    
    - Physical distance and count
    are matter
    

    View Slide

28. VMs on Cloud
    
    
    - Using computing power of
    multiple instances in a physical
    
    
    - Scale-up and down instantly
    
    
    - Physical distance is still matter
    but not count
    

    View Slide

29. Function / Lambda
    
    
    - Using computing power on centralized
    functions on cloud
    
    
    - Computing and virtualization is becoming
    more granular
    
    
    - Serverless
    
    
    - Physical distance is still matter but not count
    

    View Slide

30. Edge
    
    
    - Using computing power on decentralized
    compact functions on each Edges
    
    
    - Ultra-low latency to function and
    resources
    
    
    - Physical distance and count doesn't
    matter
    

    View Slide