Chain of Responsibility pattern in LLD - Scalability & System Analysis

Scalability Analysis - Chain of Responsibility pattern

Growth Table: Chain of Responsibility Pattern

Users / Requests	What Changes?
100 requests/sec	Single chain instance handles requests sequentially; low latency; simple setup.
10,000 requests/sec	Chain instances may become CPU-bound; latency increases; need parallel chains or load balancing.
1,000,000 requests/sec	Single server insufficient; multiple servers with replicated chains; distributed load balancing required.
100,000,000 requests/sec	Global distributed system; chains partitioned by request type; caching and asynchronous processing essential.

First Bottleneck

The first bottleneck is the processing capacity of the chain handlers. As requests increase, the sequential nature of the chain causes delays because each handler must process or pass the request. CPU and memory limits on the server running the chain break first.

Scaling Solutions

• Horizontal scaling: Run multiple chain instances on different servers to share load.
• Load balancing: Distribute incoming requests evenly across chain instances.
• Parallel chains: Partition requests by type or category so different chains handle different requests.
• Caching: Cache results of handlers that produce repeatable outputs to avoid redundant processing.
• Asynchronous processing: Use queues to decouple request reception from processing, smoothing spikes.

Back-of-Envelope Cost Analysis

• At 10,000 requests/sec, assuming each handler takes 1ms, a single chain can handle ~1000 requests/sec; need ~10 servers.
• Storage is minimal unless handlers log extensively; estimate 1KB per request -> 10MB/sec at 10K req/sec.
• Network bandwidth depends on request/response size; for 1KB each, 10MB/sec inbound and outbound at 10K req/sec.

Interview Tip

Start by explaining the chain's sequential processing and its impact on latency. Identify the handler processing as the bottleneck. Then discuss horizontal scaling and load balancing. Mention caching and asynchronous queues as optimizations. Always relate solutions to the bottleneck you identified.

Self Check

Your chain handles 1000 requests per second. Traffic grows 10x to 10,000 requests per second. What do you do first?

Answer: Add more chain instances and use load balancing to distribute requests, because a single chain cannot process 10,000 requests per second sequentially.

Key Result

The Chain of Responsibility pattern scales by adding parallel chain instances and load balancing requests, as the sequential processing in a single chain becomes the first bottleneck at higher traffic.

Practice

1. What is the main purpose of the Chain of Responsibility pattern in system design?

easy

A. To pass a request along a chain of handlers until one handles it

B. To create multiple copies of an object

C. To ensure only one instance of a class exists

D. To define a family of algorithms and make them interchangeable

Solution

1. Step 1: Understand the pattern's behavior

   The Chain of Responsibility pattern allows a request to be passed along a chain of objects until one can handle it.

2. Step 2: Compare options with the pattern's purpose

   The options describing singleton (one instance), prototype (object copies), and strategy (interchangeable algorithms) refer to other design patterns, not Chain of Responsibility.

3. Final Answer:

   To pass a request along a chain of handlers until one handles it -> Option A

4. Quick Check:

   Chain of Responsibility = pass request along chain [OK]

Hint: Remember: Chain passes request until handled [OK]

Common Mistakes:

• Confusing with Singleton or Strategy patterns
• Thinking it creates object copies
• Assuming it handles all requests at once

2. Which of the following is the correct way to link handlers in a Chain of Responsibility pattern?

easy

A. Handlers are linked using a global static list

B. Handlers are independent and do not reference each other

C. Handlers communicate through a shared database

D. Each handler holds a reference to the next handler in the chain

Solution

1. Step 1: Recall how handlers are connected

   In Chain of Responsibility, each handler has a reference to the next handler to pass the request along.

2. Step 2: Evaluate other options

   Global static lists, shared databases, and independent handlers describe unrelated or incorrect linking methods that contradict the chain concept.

3. Final Answer:

   Each handler holds a reference to the next handler in the chain -> Option D

4. Quick Check:

   Handler links = next handler reference [OK]

Hint: Handlers link by referencing the next handler [OK]

Common Mistakes:

• Using global lists instead of direct references
• Assuming handlers are independent
• Confusing with event broadcasting

3. Consider this simplified handler chain code snippet:

class Handler:
    def __init__(self, successor=None):
        self.successor = successor
    def handle(self, request):
        if self.can_handle(request):
            return f"Handled {request}"
        elif self.successor:
            return self.successor.handle(request)
        else:
            return "Not handled"
    def can_handle(self, request):
        return False

class ConcreteHandlerA(Handler):
    def can_handle(self, request):
        return request == 'A'

class ConcreteHandlerB(Handler):
    def can_handle(self, request):
        return request == 'B'

chain = ConcreteHandlerA(ConcreteHandlerB())
print(chain.handle('B'))

What is the output of this code?

medium

A. Handled B

B. Handled A

C. Not handled

D. Error

Solution

1. Step 1: Trace the request through the chain

   The request 'B' is first checked by ConcreteHandlerA, which returns False for can_handle('B'). It passes the request to ConcreteHandlerB.

2. Step 2: ConcreteHandlerB handles the request

   ConcreteHandlerB's can_handle('B') returns True, so it returns "Handled B".

3. Final Answer:

   Handled B -> Option A

4. Quick Check:

   Request 'B' handled by second handler [OK]

Hint: Request passes chain until a handler returns true [OK]

Common Mistakes:

• Assuming first handler handles all requests
• Confusing return values
• Missing the chain passing logic

4. Given the following code snippet, identify the bug that breaks the Chain of Responsibility pattern:

class Handler:
    def __init__(self, successor=None):
        self.successor = successor
    def handle(self, request):
        if self.can_handle(request):
            return f"Handled {request}"
        else:
            return "Not handled"
    def can_handle(self, request):
        return False

class ConcreteHandler(Handler):
    def can_handle(self, request):
        return request == 'X'

chain = ConcreteHandler()
print(chain.handle('Y'))

What is the main issue here?

medium

A. The successor is not initialized properly

B. The can_handle method always returns True

C. The handler does not pass the request to the successor

D. The handle method has a syntax error

Solution

1. Step 1: Analyze the handle method logic

   The handle method checks can_handle; if False, it returns "Not handled" immediately without passing to successor.

2. Step 2: Identify missing chain passing

   It should call self.successor.handle(request) if successor exists, but this is missing.

3. Final Answer:

   The handler does not pass the request to the successor -> Option C

4. Quick Check:

   Missing successor call breaks chain [OK]

Hint: Always pass request to successor if not handled [OK]

Common Mistakes:

• Forgetting to call successor.handle()
• Assuming can_handle always returns True
• Ignoring successor initialization

5. You are designing a logging system using the Chain of Responsibility pattern. You want to handle logs of different severity: DEBUG, INFO, WARNING, ERROR. Each handler should process logs at its level and pass higher severity logs down the chain. Which design best fits this requirement?

hard

A. Each handler processes only its level and passes all logs to the next handler

B. Each handler processes logs at its level and passes only higher severity logs to the next handler

C. Each handler processes all logs regardless of severity and stops the chain

D. Each handler processes logs randomly without order

Solution

1. Step 1: Understand the logging severity flow

   Logs should be handled at their level, and higher severity logs should continue down the chain for further handling.

2. Step 2: Evaluate options for correct chain behavior

   Each handler processes logs at its level and passes only higher severity logs to the next handler matches this: handlers process their level and pass higher severity logs onward. Each handler processes only its level and passes all logs to the next handler passes all logs regardless, which is inefficient. Each handler processes all logs regardless of severity and stops the chain stops chain prematurely. Each handler processes logs randomly without order is random and incorrect.

3. Final Answer:

   Each handler processes logs at its level and passes only higher severity logs to the next handler -> Option B

4. Quick Check:

   Process level, pass higher severity [OK]

Hint: Process own level, pass higher severity logs down [OK]

Common Mistakes:

• Passing all logs without filtering
• Stopping chain too early
• Ignoring severity order