Module 5: Threadpools & Asynchronous Programming

Question 1

Why can singular threads create an issue?

Answer 1

• Threads consume resources during creation/initialization/context switching
• Number of threads increases = each thread does less actual work = decreased efficiency
• Overhead of threads surpasses system’s capacity = system crash

Question 2

How do thread pools work?

Answer 2

• Thread pool manages a pool of worker threads & assigns them to execute the work items/tasks.

Question 3

What are asynchronous calls?

Answer 3

• Programming mechanism that allows a function to be executed ina non-blocking manner
• Basically, program doesn’t wait for the task to complete before moving on to execute the next line of code
• EX use: Web servers & GUI, improves performance & responsiveness.

Question 4

Thread pool queues

Answer 4

• Crucial component of thread pool design
• Tasks are added to a work queue by application, can be viewed as a shared resource that is associated by the worker threads

Question 5

How do the thread pool queues improve performance?

Answer 5

• Thread finds available task = picks it up & executes it
• Eliminates need to create & destroy threads continuously

Question 6

Dynamic & static thread pool queues

Answer 6

• Fixed-size queue: limits nr of tasks that can be queued. Maximum capacity = new tasks rejected until a slot becomes available
• Dynamic queue: Automatically adjusts its size based on the workload

Question 7

How can thread pool queues ensure load balance?

Answer 7

• Beneficial for load balancing since they help distribute tasks evenly among worker threads
• No thread is overloaded
• Results = improved performance & reduced latency

Question 8

The downsides of using a single thread for clients’ requests, or a seperate thread for each request?

Answer 8

One thread: Takes too long
New threads: Overhead

Question 9

How can a thread pool with a limited nr of worker threads be effective in handling requests?

Answer 9

• Request arrives: server adds it to queue of pending requests
• Worker threads periodically check queue for new request. If it exists, it picks it up. When done, returns to the pool.
• Results:
• No running outta system resources
• Improved responsiveness
• No complex thread management for the programmer

Question 10

Creating a thread pool in C# using a lambda expression function

Answer 10

public void Create()
{
ThreadPool.QueueUserWorkItem((state) =>
{
Console.WriteLine(“A thread pool has been
made”);
});

Question 11

Disadvantages of thread pools

Answer 11

1. Lack of control over threads. Never know when a thread has started, finished, etc. Difficult to debug.
2. Limited scalability. Pools have fixed number of threads, not scalable for apps with large number of threads. Too many threads = app crash.
3. Potential for thread starvation. Task waits for available thread to execute, but all threads are busy. Delay in task execution = decreased performance.
4. Complex programming. Requires careful programming => development process more difficult & time-consuming.
5. Increased memory usage. Increase memory usage if nr of threads is too high. Each thread requires own stack & other resources, can quickly add up in a large thread pool.

Question 12

When to use thread pools?

Answer 12

1. Apps that create & destroy many threads
2. Apps with many small tasks to be done asynchronously
3. Apps that process independent work items in the background & in parallel

Question 13

When not to use thread pools?

Answer 13

1. Apps with tasks having varying execution times
2. Apps that require fine-grained control over threads or require specific thread attributes

Question 14

What is asynchronous programming?

Answer 14

• Type of parallel programming
• Enables unit of work to execute seperately from the main application thread
• Task finished = notifies caller thread, whether successful or failed.
• Provides a nin-block, event-driven model that enhances responsiveness & performance

Question 15

What is the non-blocking model?

Answer 15

• Can use the same thread to process many tasks, not needing to complete one before moving on to the next
• Improves performance by utilixing thread to execute many tasks concurrently
• (Gas station example)

Question 16

Single-threaded sync code

Answer 16

• Sync sequential code = executes in deterministic order, produce the same results for the same input everytime.
• Programmer has complete control over operations
• Easy to read, straightforward BUT… =>
• May be slow = unresponsive app

Question 17

Multiple-threaded synchronous code

Answer 17

• Each task executes on a seperate thread => tasks run concurrently or in parallel
• Less control over execution order of tasks
• Requires effective communication & coordination between threads

Question 18

Asynchronous model

Answer 18

• Allows tasks to execute independently & concurrently, without blocking main thread of application
• Not necessarily have seperate threads for each task. Tasks can be managed by a single thread or small pool of threads.

Question 19

What are some benefits of asynchronous model?

Answer 19

• Programmer has more control over the execution order of tasks
• greater flexibility in scheduling & prioritizing tasks = improve application performance & responsiveness

Question 20

When should you use single-threaded async code?

Answer 20

• Apps with large number of tasks, esp that require little communication btw tasks
• Only blocks when no task can make progress = referred to as non-blocking program. Reducing waiting time = improve app performance.
• Good resource utilization & efficiency = reduces context switching & minimizes memory consumption

Question 21

Practical example of async

Answer 21

• Graphical user interface (GUI) needs to be responsive
• Too many threads = decrease responsiveness. Not good for user experience.
• Async with a single thread = GUI maintain high level of responsiveness = better user experience

Question 22

Multi-threaded async code

Answer 22

• Execution of multiple instructions on multiple cores/CPU simultaneously
• Utilizes potentia of multi-core machines. Allows highly concurrent & true parallelism = super performance!!

Question 23

Async and thread pools?

Answer 23

• Thread pools used in async programming for efficient use of system resources
• Allows overhead of creating & managing many threads
• Useful for apps with short-lived tasks

Question 24

Defining tasks

Answer 24

• Task in async programming = unit of work that needs to be executed asynchronously.
• Can represent any operation that needs to be executed, eg reading data to a file

Question 25

Promises/futures

Answer 25

• Tasks return a promise or a future object
• Caller of the method can use this object to track the progress of the task & retrive its result when completed

Question 26

Tasks in C#

Answer 26

• Task Class
• async = marks method as asynchronous, perform work without blocking the calling thread
• await keywords = wait for completion of other tasks
• Task.WhenAll = waits for all provided tasks to complete
• Task.WhenAny = waits for the first task to complete
• CancellationToken class = cancel asynchronous operation before completion