Concurrency vs Parallelism

A good code is one which uses the system resources efficiently which means not over utilizing the resources as well as not under utilizing by leaving them idle. One of the main features of Python3 is its asynchronous capabilities. After getting to know what is concurrency and why concurrency is not parallelism. We will see how can we leverage the asyncio library's functionalities to design and run our code concurrently, use our resources efficiently and finish the tasks blazingly fast.

Transcript

1. CONCURRENCY
   VS
   PARALLELISM
   Narendran
   @DudeWhoCode
   

   View Slide

2. $ WHO AM I
   Naren
   Senior Software Consultant
   Tarka Labs

   

   Cycling | Travelling | Blogging
   @DudeWhoCode
   

   View Slide

3. @DudeWhoCode

   blog.dudewho.codes
   

   View Slide

4. SYNC VS ASYNC
   @DudeWhoCode
   

   View Slide

5. CONCURRENT VS PARALLEL
   @DudeWhoCode
   

   View Slide

6. CONCURRENT VS PARALLEL
   @DudeWhoCode
   

   View Slide

7. RECAP
   • Sync: Blocking operations.
   • Async: Non blocking operations.
   • Concurrency: Making progress together.
   • Parallelism: Making progress in parallel.
   @DudeWhoCode
   

   View Slide

8. CONCURRENCY IS NOT PARALLELISM
   @DudeWhoCode
   

   View Slide

9. CONCURRENCY IS NOT PARALLELISM
   @DudeWhoCode
   

   View Slide

10. @DudeWhoCode
    

    View Slide

11. @DudeWhoCode
    

    View Slide

12. @DudeWhoCode
    

    View Slide

13. @DudeWhoCode
    

    View Slide

14. @DudeWhoCode
    

    View Slide

15. @DudeWhoCode
    

    View Slide

16. @DudeWhoCode
    

    View Slide

17. @DudeWhoCode
    

    View Slide

18. CONCURRENCY IN PYTHON
    3
    @DudeWhoCode
    

    View Slide

19. ASYNCIO
    • Python 3 standard library to write single-threaded
    concurrent code
    • Introduced in python 3.4
    • Inspired from “twisted” library
    @DudeWhoCode
    

    View Slide

20. ASYNCIO
    • Event Loops
    • Coroutines
    • Futures
    @DudeWhoCode
    

    View Slide

21. EVENT LOOP
    • Manages and distributes the execution of different
    tasks.
    •Responsible for registering the tasks and distributing
    flow of control between them
    @DudeWhoCode
    

    View Slide

22. COROUTINES
    •Special functions capable of releasing the flow of
    control back to the event loop.
    •Similar to python generators
    •A coroutine is scheduled using an event loop
    @DudeWhoCode
    

    View Slide

23. FUTURES
    •Objects that represent the result of a task.
    •Task can be completed or pending.
    •Object may be exceptions too.
    @DudeWhoCode
    

    View Slide

24. FLOW OF EXECUTION
    @DudeWhoCode
    

    View Slide

25. @DudeWhoCode
    

    View Slide

26. DIVING DEEP INTO CONCURRENCY…
    @DudeWhoCode
    

    View Slide

27. CONTEXT SWITCH
    @DudeWhoCode
    

    View Slide

28. @DudeWhoCode
    

    View Slide

29. BLOCKING TASKS
    @DudeWhoCode
    

    View Slide

30. @DudeWhoCode
    

    View Slide

31. ORDER OF EXECUTION
    @DudeWhoCode
    

    View Slide

32. @DudeWhoCode
    

    View Slide

33. SYNC VS ASYNC HTTP CALLS
    @DudeWhoCode
    

    View Slide

34. @DudeWhoCode
    

    View Slide

35. @DudeWhoCode
    

    View Slide

36. RETURN WHEN
    @DudeWhoCode
    

    View Slide

37. @DudeWhoCode
    

    View Slide

38. CANCEL FUTURE
    @DudeWhoCode
    

    View Slide

39. @DudeWhoCode
    

    View Slide

40. WAIT TIMEOUT
    @DudeWhoCode
    

    View Slide

41. @DudeWhoCode
    

    View Slide

42. CONCURRENCY VS PARALLELISM
    @DudeWhoCode
    

    View Slide

43. PARALLEL HTTP CALL
    @DudeWhoCode
    

    View Slide

44. @DudeWhoCode
    

    View Slide

45. CONCURRENT HTTP CALL
    @DudeWhoCode
    

    View Slide

46. @DudeWhoCode
    

    View Slide

47. CONCURRENCY VS PARALLELISM

    BENCHMARK VIDEO
    @DudeWhoCode
    

    View Slide

48. @DudeWhoCode
    

    View Slide

49. HOW CONCURRENCY IS DIFFERENT FROM
    MULTI-THREADING?
    @DudeWhoCode
    

    View Slide

50. THREADING VS ASYNC
    • Async : Mostly single threaded, You decide when
    a piece of code can take back control using await
    • Threading : Python scheduler takes care of
    context switch and it may lose control anytime.
    @DudeWhoCode
    

    View Slide

51. SUMMARY
    • Sync: Blocking operations.
    • Async: Non blocking operations.
    • Concurrency: Creating subtasks, scheduling subtasks,
    switching between them, making progress together.
    • Parallelism: Creating subtasks, utilize CPU core,
    making progress in parallel.
    @DudeWhoCode
    

    View Slide

52. SUMMARY
    • Python 3, asyncio, aiohttp, aiofiles
    • Eventloops, co-routines, futures
    @DudeWhoCode
    

    View Slide

53. SUMMARY
    • context switching
    • sync vs async http calls
    • no order of execution
    • return_when
    • cancel pending future
    • wait timeout
    • parallel vs concurrent http calls
    @DudeWhoCode
    

    View Slide

54. REFERENCES
    • AsyncIO for the Working Python Developer (Hackernoon)

    - Yeray Diaz
    • A guide to asynchronous programming in Python with asyncio
    (FreeCodeCamp)

    - Roman Gaponov
    @DudeWhoCode
    

    View Slide

55. THANK YOU,
    NAREN
    @DudeWhoCode
    Slides:

    www.dudewho.codes/talks
    @DudeWhoCode
    

    View Slide