Making Lambda simpler for data scientists

Presented at AWS Community Day Bengaluru 2019

Transcript

1. Nabarun Pal (@theonlynabarun)
   Making Lambda Simpler
   for Data Scientists
   

   View Slide

2. About Me
   ● Platform Engineer at rorodata
   ● Optimizing development time through simple abstractions/tooling
   ● Venturing into Container Orchestration and Serverless Computing
   ● Contributor to the Kubernetes ecosystem
   

   View Slide

3. Outline
   ● Genesis
   ● Present Constructs in Python - Threads and Processes
   ● Kubernetes
   ● Serverless
   ● The Abstraction
   ● Requirements
   ● API
   ● Internals
   ● Demo
   ● Performance Metrics
   ● Current Limitations
   ● Future Goals
   

   View Slide

4. Genesis
   

   View Slide

5. Multithreading
   Pros
   ● Lightweight
   ● Shared state between multiple threads
   ● Works flawlessly for I/O-bound applications
   Cons
   ● Subject to Global Interpreter Lock
   ● Context switching overhead
   ● Code prone to race conditions
   ● Does not work for CPU-bound tasks
   

   View Slide

6. Multiprocessing
   Pros
   ● Isolation of memory space
   ● Leverages multiples processors & cores
   ● GIL limitations don’t apply
   ● Synchronization primitives like locks are mandatory unless sharing data
   ● Works well for CPU-bound tasks
   Cons
   ● Sharing data between processes is a little bit complicated
   ● Bulky memory footprint
   ● Definite scaling
   

   View Slide

7. Can Kubernetes help?
   Pros
   ● Abstracts out infrastructure
   ● Simple interface
   ● Can scale based on workload
   Cons
   ● Layer on top of VM’s - Slow to scale up/down
   ● Autoscaling is not a core functionality
   ● Needs dedicated time to manage
   

   View Slide

8. What about Serverless?
   ● Zero Infrastructure Management
   ● Near Infinite Scaling
   ● High Availability
   ● No Idle Resources
   ● Suitable for short-lived workloads
   

   View Slide

9. All problems in computer science can be
   solved by another level of indirection
   David Wheeler
   

   View Slide

10. LambdaPool
    - “The Indirection”
    

    View Slide

11. Requirements
    ● Minimum overhead on users
    ● Simple way to create, delete, list and update lambda functions
    ● Coherent ways to invoke the lambda function
    ● Easy to use interface
    

    View Slide

12. Features
    ● CLI to create, list, update and delete functions
    ● Support for specifying function layers and list the layers used for each functions
    ● LambdaPool interface
    ● LambdaExecutor Interface
    

    View Slide

13. Installation
    $ pip install --user https://lambdapool-releases.s3.amazonaws.com/lambdapool-0.9.7.tar.gz
    

    View Slide

14. CLI
    

    View Slide

15. LambdaPool
    ● Implements the same interface as ThreadPool and ProcessPool
    

    View Slide

16. LambdaExecutor
    ● Implements the same interface as ThreadPoolExecutor and ProcessPoolExecutor
    

    View Slide

17. Demo
    

    View Slide

18. Internals - Creating/Updating a Function
    

    View Slide

19. Internals - Invoking a function
    

    View Slide

20. Benefits
    ● Compute Time
    ● Compute Costs
    ● Developer Time
    

    View Slide

21. Current Limitations
    ● Serialization of the payload is being a hurdle
    ● Decoupling between function provisioning and invocation
    ● Size of execution environment
    Inherent to Serverless
    ● Cold start issues
    ● Additional Network Overhead
    ● Not suitable for long running workloads
    ● Troubleshooting is hard
    ● Local testing
    

    View Slide

22. Future Goals
    ● Distribute lambdapool through PyPI
    ● Permissions management system
    ● System to fetch execution logs
    ● Better layer management
    ● Make the function update process intelligent
    

    View Slide

23. Conclusions
    https://github.com/rorodata/lambdapool
    

    View Slide

24. Thank You!
    Feedback: https://bit.ly/lambdapoolfeedback Contact Us: opensource@rorodata.com
    

    View Slide