System Design (HLD)Separation of Concerns

Separation of Concerns

LevelIntermediate

Duration75 mins

TopicSeparation of Concerns

1 / 4

Identifying Concerns

The Essence of Software Decomposition

Every software system, regardless of its size or complexity, is a tapestry woven from distinct threads of functionality. A web application simultaneously handles user authentication, data validation, business logic, persistence, logging, error handling, and presentation. Each of these represents a different concern—a distinct area of responsibility within the system.

The ability to recognize and articulate these concerns is the first step toward building software that remains comprehensible and malleable throughout its lifecycle. Without this recognition, systems become tangled masses where every change requires understanding everything at once, and modifying one feature inadvertently breaks another.

What You Will Learn

By the end of this page, you will understand what constitutes a 'concern' in software design, develop techniques for identifying concerns in any system, and learn to distinguish between different types of concerns. This foundational skill enables every subsequent principle of separation—you cannot separate what you cannot see.

What Is a Concern?

In software engineering, a concern is any piece of interest or focus in a program. More precisely, it is a distinct aspect of functionality that can be identified, discussed, and potentially isolated from other aspects. The term was popularized by Edsger W. Dijkstra in his seminal 1974 paper "On the role of scientific thought," where he introduced the concept of separation of concerns as a fundamental principle of program organization.

Dijkstra's Original Insight:

"Let me try to explain to you, what to my taste is characteristic for all intelligent thinking. It is, that one is willing to study in depth an aspect of one's subject matter in isolation for the sake of its own consistency, all the time knowing that one is occupying oneself only with one of the aspects. We know that a program must be correct and we can study it from that viewpoint only; we also know that it should be efficient and we can study its efficiency on another day, so to speak. In another mood we may ask ourselves whether, and if so: why, the program is desirable. But nothing is gained—on the contrary!—by tackling these various aspects simultaneously. It is what I sometimes have called 'the separation of concerns', which, even if not perfectly possible, is yet the only available technique for effective ordering of one's thoughts."

This insight remains as powerful today as when it was written. A concern represents a coherent area of interest that can be reasoned about independently—even though in the final system, all concerns must work together harmoniously.

Essential Characteristics of a Concern

•Coherence — A concern represents a unified, logically related set of behaviors or data. Everything within the concern 'belongs together' in a meaningful way.
•Identifiability — A concern can be named and discussed. If you can't articulate what a concern is about, it may not be a genuine concern but rather an undefined mixture.
•Distinct Purpose — A concern serves a specific purpose that differs from other concerns. While concerns may interact, each addresses a different fundamental question or responsibility.
•Potential for Isolation — At least in principle, a concern could be developed, tested, or modified without deep entanglement with other concerns (though full isolation isn't always practical).
•Stable Boundary — Over time, the boundary of a well-identified concern remains relatively stable, even as implementation details evolve.

Concerns Are Not Always Obvious

Identifying concerns is an analytical skill that develops with experience. A junior developer might see a monolithic 'handle user request' function. A senior developer sees authentication, authorization, validation, business logic, persistence, and response formatting—each a distinct concern waiting to be recognized.

Categories of Concerns

Concerns manifest at different levels and serve different purposes within software systems. Understanding these categories helps you systematically analyze any system and ensure you haven't overlooked important aspects.

Functional vs Non-Functional Concerns:

The most fundamental distinction is between concerns that define what the system does versus how well it does it:

Functional vs Non-Functional Concerns
Category	Definition	Examples	Typical Questions
Functional Concerns	What the system does—its features and capabilities	User registration, payment processing, order fulfillment, report generation	What actions can users perform? What data transformations occur?
Non-Functional Concerns	How well the system performs its functions—quality attributes	Performance, security, reliability, scalability, usability, maintainability	How fast? How secure? How available? How easy to modify?

Domain Concerns vs Technical Concerns:

Another critical distinction separates concerns related to the problem domain from concerns related to the technical solution:

Domain Concerns

•Business rules — How the domain works (e.g., 'orders over $100 get free shipping')
•Domain entities — Core concepts like Customer, Order, Product, Account
•Domain services — Operations meaningful in the business context (e.g., 'process refund')
•Workflows — Business processes that span multiple operations
•Policies — Business decisions encoded as logic (pricing, eligibility, etc.)

Technical Concerns

•Persistence — How data is stored and retrieved (databases, files)
•Communication — How components exchange data (HTTP, messaging)
•Infrastructure — Deployment, scaling, monitoring capabilities
•Presentation — How information is displayed to users (UI/UX)
•Integration — How external systems are connected and managed

The Volatility Principle

Domain concerns and technical concerns typically change for different reasons and at different rates. Business rules evolve with market conditions; infrastructure concerns change with technology choices. Separating them means that replacing your database doesn't require rewriting your business logic, and changing a business rule doesn't require understanding your caching strategy.

Core Application Concerns

While every application has unique functional requirements, certain concerns appear so consistently across software systems that they form a common vocabulary for engineers. Recognizing these patterns accelerates your ability to analyze new systems and communicate with other developers.

The Universal Concerns:

Consider a typical web application processing a user request. Even in a simple CRUD operation, numerous concerns are involved:

concerns-in-action.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
// A seemingly simple function that actually addresses many concerns
async function createOrder(request: CreateOrderRequest): Promise<OrderResponse> {
    // CONCERN: Input Validation
    // Ensuring data meets structural and semantic requirements
    validateOrderRequest(request);
    
    // CONCERN: Authentication
    // Verifying the identity of the requester
    const user = await authenticateRequest(request.token);
    
    // CONCERN: Authorization
    // Determining if the authenticated user can perform this action
    await authorizeOrderCreation(user, request);
    
    // CONCERN: Business Logic / Domain Rules
    // Applying domain-specific calculations and validations
    const orderTotal = calculateOrderTotal(request.items);
    const discounts = applyDiscountRules(orderTotal, user);
    const finalTotal = orderTotal - discounts;
    
    // CONCERN: Inventory Management
    // Checking and reserving stock
    await reserveInventory(request.items);
    
    // CONCERN: Persistence
    // Storing data durably
    const order = await orderRepository.save({
        userId: user.id,
        items: request.items,
        total: finalTotal,
        status: 'pending'
    });
    
    // CONCERN: External Integration
    // Communicating with payment provider
    await paymentService.authorize(order.id, finalTotal);
    
    // CONCERN: Event Publishing / Notification
    // Informing other parts of the system
    await eventBus.publish('order.created', { orderId: order.id });
    
    // CONCERN: Logging / Observability
    // Recording what happened for debugging and analytics
    logger.info('Order created', { orderId: order.id, userId: user.id });
    
    // CONCERN: Response Formatting / Presentation
    // Preparing the output for the caller
    return formatOrderResponse(order);
}

The code above is instructive but problematic from a separation of concerns perspective. All these concerns are intertwined in a single function, making it difficult to:

Test any single concern in isolation
Modify one concern without risking others
Reuse individual concerns in other contexts
Understand what the function does at a glance

Recognizing these concerns is the first step. Later in this module, we'll learn techniques to isolate them.

Catalog of Common Concerns

•Input Handling — Parsing, deserializing, and validating incoming data before it enters the system
•Authentication — Verifying identity claims (who is making the request)
•Authorization — Enforcing access control (what are they allowed to do)
•Business Logic — Core domain rules and calculations that define the system's purpose
•Data Access — Reading from and writing to persistent storage
•Caching — Storing computed results to avoid redundant work
•Transaction Management — Ensuring atomicity and consistency of state changes
•Error Handling — Detecting, categorizing, and responding to failures
•Logging — Recording events for debugging, auditing, and analytics
•Metrics / Monitoring — Measuring system behavior for operational visibility
•Event Publishing — Notifying other components or systems of state changes
•Output Formatting — Transforming internal data into appropriate output formats

Techniques for Identifying Concerns

Identifying concerns is part analysis, part intuition—and it improves dramatically with practice. The following techniques provide systematic approaches to uncovering concerns in any system.

Technique 1: The 'Why' Chain

For any piece of code or functionality, ask 'Why does this exist?' repeatedly until you reach a fundamental purpose. Each distinct answer often reveals a separate concern.

Example: Why does this line validate the email format? → To ensure data quality (Input Validation concern) → Why ensure data quality? To prevent errors later (Error Prevention concern) → Why prevent errors? To maintain system integrity (Data Integrity concern)

Each level reveals a different aspect that might warrant separate consideration.

why-chain-analysis.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// Original mixed concern code
function processPayment(payment: Payment) {
    // Why? To ensure the amount is valid (VALIDATION)
    if (payment.amount <= 0) throw new Error("Invalid amount");
    
    // Why? To track what happened (LOGGING)
    console.log(`Processing payment: ${payment.id}`);
    
    // Why? To check user permission (AUTHORIZATION)
    if (!payment.user.canMakePayments) throw new Error("Not authorized");
    
    // Why? To persist the payment record (PERSISTENCE)
    database.save(payment);
    
    // Why? To communicate with payment provider (INTEGRATION)
    stripeClient.charge(payment.amount, payment.user.card);
    
    // Why? To inform other systems (EVENTING)
    eventBus.emit('payment.processed', payment);
}
 
// Each "Why?" answer points to a distinct concern that could be separated

Technique 2: The Stakeholder Lens

Different stakeholders care about different aspects of a system. Viewing the system through each stakeholder's perspective reveals concerns that might otherwise be invisible:

End Users see features, usability, performance
Business Owners see revenue, compliance, competitive advantage
Operations Teams see deployability, monitoring, incident response
Developers see maintainability, testability, architectural clarity
Security Teams see vulnerabilities, access control, audit trails
Data Teams see data quality, analytics, privacy compliance

Each stakeholder perspective illuminates different concerns.

The Change Driver Test

Ask: 'What different forces might cause this code to change?' Each distinct force typically represents a distinct concern. If marketing requirements, database schema changes, and regulatory updates could all modify the same code, you have multiple concerns entangled. Well-separated concerns change for a single reason each.

Technique 3: The Verbal Description Test

Describe what a component does in plain English. If you find yourself using the word 'and' frequently, you're likely describing multiple concerns:

Poor (multiple concerns): "This class handles user registration and sends welcome emails and logs the new user and validates email uniqueness."

Better (focused concerns):

"This class orchestrates the user registration process."
"This class sends templated emails to users."
"This class validates email uniqueness against the database."
"This class logs significant user events."

Each simple description points to a single concern.

Concern Identification Checklist

•Can you name this functionality in 2-3 words? If not, it may be multiple concerns.
•Would this functionality make sense in a different application? Universal applicability often indicates a separable concern.
•Does this functionality change for a single, identifiable reason? Multiple change drivers suggest multiple concerns.
•Can you imagine testing this aspect independently? Testability indicates separability.
•Would a different team member naturally own this? Organizational ownership often aligns with concerns.
•Does this functionality have its own vocabulary or domain terms? Distinct terminology suggests a distinct concern.

Granularity of Concerns

Concerns exist at multiple levels of granularity. A common mistake is applying separation of concerns at too fine or too coarse a level. Understanding the hierarchy of concerns helps you make appropriate decisions about where to invest in separation.

Concern Hierarchy:

Concerns can be analyzed at progressively finer levels of detail:

Levels of Concern Granularity
Level	Scope	Examples	Typical Separation Mechanism
System	Entire product or platform	E-commerce vs CMS vs Analytics	Separate systems, APIs, deployments
Subsystem/Bounded Context	Major functional area	Ordering, Inventory, Payments, Shipping	Microservices, modules, packages
Component/Service	Coherent functional unit	Order validation, price calculation, notification	Classes, services, functions
Aspect	Cross-cutting behavior	Logging, security, transaction management	Aspects, middleware, decorators
Method/Function	Single operation	Calculate discount, format date, validate email	Functions, private methods

granularity-examples.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
// SYSTEM LEVEL: Entire products/platforms
// - OrderingSystem
// - InventorySystem
// - PaymentGateway
 
// SUBSYSTEM LEVEL: Bounded contexts within a product
namespace OrderingContext {
    // Everything related to orders
    class OrderService { /* ... */ }
    class OrderRepository { /* ... */ }
    class OrderEvents { /* ... */ }
}
 
namespace PaymentContext {
    // Everything related to payments  
    class PaymentService { /* ... */ }
    class PaymentRepository { /* ... */ }
    class PaymentEvents { /* ... */ }
}
 
// COMPONENT LEVEL: Services within a bounded context
class OrderService {
    constructor(
        private validator: OrderValidator,      // Validation concern
        private calculator: PriceCalculator,    // Calculation concern
        private repository: OrderRepository,    // Persistence concern
        private events: OrderEventPublisher     // Event concern
    ) {}
}
 
// ASPECT LEVEL: Cross-cutting concerns
// Applied via decorators/middleware
@Transactional()
@Logged()
@Cached({ ttl: 300 })
async getOrderById(id: string): Promise<Order> { ... }
 
// METHOD LEVEL: Single operations
class PriceCalculator {
    calculateBasePrice(items: Item[]): Money { /* focused logic */ }
    applyDiscounts(price: Money, discounts: Discount[]): Money { /* focused logic */ }
    applyTax(price: Money, region: Region): Money { /* focused logic */ }
}

Finding the Right Granularity

Too fine-grained separation creates fragmentation—dozens of tiny classes that individually make no sense. Too coarse-grained separation fails to achieve any real separation—a few massive classes with everything tangled together. The right granularity is where each unit has a clear, coherent purpose and changes for a single reason. This is art as much as science.

Concerns Across Programming Paradigms

The concept of concerns transcends any specific programming paradigm, but different paradigms offer different mechanisms for expressing and separating concerns.

Object-Oriented Programming:

OOP provides classes and interfaces as the primary mechanism for separating concerns. Each class ideally represents a single concern, and interfaces define the contracts between concerns:

oop-concerns.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
// OOP: Classes encapsulate concerns
interface OrderRepository {
    save(order: Order): Promise<void>;      // Persistence concern
    findById(id: string): Promise<Order>;
}
 
interface PaymentProcessor {
    process(amount: Money): Promise<PaymentResult>;  // Payment concern
}
 
interface NotificationService {
    notify(user: User, message: Message): Promise<void>;  // Notification concern
}
 
// Concerns are composed via dependency injection
class OrderService {
    constructor(
        private repository: OrderRepository,
        private payment: PaymentProcessor,
        private notifications: NotificationService
    ) {}
    
    async createOrder(order: Order): Promise<void> {
        // This method orchestrates but doesn't implement the concerns
        await this.repository.save(order);
        await this.payment.process(order.total);
        await this.notifications.notify(order.user, orderConfirmation(order));
    }
}

Functional Programming:

FP separates concerns through pure functions and composition. Each function addresses a single transformation, and concerns are separated by avoiding side effects:

fp-concerns.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
// FP: Functions represent concerns as pure transformations
// Each function handles one concern
 
// Validation concern (pure)
const validateOrder = (order: Order): Either<ValidationError[], Order> => {
    const errors = [
        validateItems(order.items),
        validateCustomer(order.customer),
        validateShipping(order.shipping)
    ].filter(isError);
    
    return errors.length > 0 ? left(errors) : right(order);
};
 
// Calculation concern (pure)  
const calculateTotal = (order: Order): OrderWithTotal => ({
    ...order,
    total: order.items.reduce((sum, item) => sum + item.price * item.quantity, 0)
});
 
// Discount concern (pure)
const applyDiscounts = (order: OrderWithTotal): OrderWithTotal => ({
    ...order,
    total: order.total * (1 - getDiscountRate(order))
});
 
// Composition: pipeline of concerns
const processOrder = pipe(
    validateOrder,
    map(calculateTotal),
    map(applyDiscounts),
    chain(persistOrder),    // Side effect isolated
    chain(sendNotification)  // Side effect isolated
);

Aspect-Oriented Programming:

AOP provides explicit mechanisms for separating cross-cutting concerns (those that span multiple components) from primary logic:

aop-concerns.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
// AOP: Cross-cutting concerns declared separately from business logic
 
// Business logic - clean and focused
class OrderService {
    async createOrder(order: Order): Promise<Order> {
        const validated = this.validator.validate(order);
        const priced = this.calculator.calculateTotal(validated);
        return this.repository.save(priced);
    }
}
 
// Cross-cutting concerns declared as aspects
@Aspect()
class LoggingAspect {
    @Before('execution(* OrderService.*(..))')
    logEntry(jp: JoinPoint) {
        console.log(`Entering ${jp.methodName} with args: ${jp.args}`);
    }
    
    @After('execution(* OrderService.*(..))')
    logExit(jp: JoinPoint) {
        console.log(`Exiting ${jp.methodName}`);
    }
}
 
@Aspect()
class TransactionAspect {
    @Around('execution(* *Repository.save(..))')
    wrapInTransaction(jp: JoinPoint) {
        const tx = this.transactionManager.begin();
        try {
            const result = jp.proceed();
            tx.commit();
            return result;
        } catch (e) {
            tx.rollback();
            throw e;
        }
    }
}

Paradigm Agnostic Principle

Regardless of paradigm, the goal remains the same: organize code so that each piece has a clear, single purpose and can be understood, modified, and tested in relative isolation. The paradigm provides the tools; the principle guides their use.

Identifying Concerns: Practical Exercises

Let's apply what we've learned to analyze real code and identify its concerns. These exercises build the analytical muscle needed for effective separation.

Exercise 1: The God Function

Examine the following function and identify all distinct concerns:

exercise-god-function.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
async function handleUserRegistration(req: Request, res: Response) {
    // What concerns can you identify here?
    const startTime = Date.now();
    
    const { email, password, name } = req.body;
    
    if (!email || !email.includes('@')) {
        return res.status(400).json({ error: 'Invalid email' });
    }
    
    if (!password || password.length < 8) {
        return res.status(400).json({ error: 'Password too short' });
    }
    
    try {
        const existing = await db.users.findOne({ email });
        if (existing) {
            return res.status(409).json({ error: 'Email already registered' });
        }
        
        const hashedPassword = await bcrypt.hash(password, 10);
        
        const user = await db.users.insert({
            email,
            password: hashedPassword,
            name,
            createdAt: new Date()
        });
        
        const token = jwt.sign({ userId: user.id }, SECRET, { expiresIn: '24h' });
        
        await sendEmail({
            to: email,
            subject: 'Welcome!',
            body: `Welcome to our platform, ${name}!`
        });
        
        await analytics.track('user.registered', { userId: user.id });
        
        console.log(`User ${email} registered in ${Date.now() - startTime}ms`);
        
        res.cookie('authToken', token, { httpOnly: true });
        res.status(201).json({ user: { id: user.id, email, name } });
        
    } catch (error) {
        console.error('Registration failed:', error);
        await alerting.notify('registration-failure', error);
        res.status(500).json({ error: 'Registration failed' });
    }
}

Concerns in the Registration Function

•Request Parsing — Extracting data from the HTTP request body
•Input Validation — Checking email format and password length
•Uniqueness Verification — Checking if email is already registered
•Password Security — Hashing passwords before storage
•User Persistence — Storing the user in the database
•Token Generation — Creating an authentication JWT
•Email Notification — Sending the welcome email
•Analytics Tracking — Recording the registration event
•Performance Logging — Timing and logging the operation
•Response Formatting — Setting cookies and JSON response
•Error Handling — Catching and responding to failures
•Alerting — Notifying operations of failures

The Separation Challenge

Notice how this 'simple' function embeds 12+ distinct concerns in roughly 50 lines. Each concern could change independently: validation rules might evolve, email templates might change, analytics providers might switch. Yet they're all tangled together. In the next pages, we'll learn techniques for isolating these concerns, making each one testable and modifiable on its own.

Summary: Identifying Concerns

Identifying concerns is the foundational skill for separation of concerns. Before you can separate, you must see. Let's consolidate what we've learned:

Key Takeaways

•A concern is a coherent area of interest — any distinct aspect of functionality that can be identified, named, and discussed independently.
•Concerns exist in categories — functional vs non-functional, domain vs technical. Each category brings different considerations.
•Common concerns recur across systems — authentication, validation, persistence, logging, and others appear in nearly every application.
•Use systematic techniques — the 'Why' chain, stakeholder lens, change driver test, and verbal description test help uncover hidden concerns.
•Granularity matters — concerns exist at system, subsystem, component, aspect, and method levels. Choose appropriate granularity for separation.
•Programming paradigms offer different tools — OOP uses classes, FP uses pure functions, AOP uses aspects. The principle transcends the mechanism.

What's Next:

Now that we can identify concerns, the next page explores Isolating Concerns—the techniques and patterns for actually achieving separation in code. We'll learn how to structure systems so that each concern lives in its own space, interacting with others through well-defined boundaries.

Page Complete

You now understand what concerns are and how to identify them systematically. This analytical skill is the foundation for all separation of concerns work—you can't separate what you can't see. In the next page, we'll learn how to translate this understanding into structured, well-organized code.

1 / 4

Loading learning content...

System Design (HLD)Separation of Concerns

Separation of Concerns

LevelIntermediate

Duration75 mins

TopicSeparation of Concerns

1 / 4

Identifying Concerns

The Essence of Software Decomposition

What You Will Learn

What Is a Concern?

Dijkstra's Original Insight:

"Let me try to explain to you, what to my taste is characteristic for all intelligent thinking. It is, that one is willing to study in depth an aspect of one's subject matter in isolation for the sake of its own consistency, all the time knowing that one is occupying oneself only with one of the aspects. We know that a program must be correct and we can study it from that viewpoint only; we also know that it should be efficient and we can study its efficiency on another day, so to speak. In another mood we may ask ourselves whether, and if so: why, the program is desirable. But nothing is gained—on the contrary!—by tackling these various aspects simultaneously. It is what I sometimes have called 'the separation of concerns', which, even if not perfectly possible, is yet the only available technique for effective ordering of one's thoughts."

Essential Characteristics of a Concern

•Coherence — A concern represents a unified, logically related set of behaviors or data. Everything within the concern 'belongs together' in a meaningful way.
•Identifiability — A concern can be named and discussed. If you can't articulate what a concern is about, it may not be a genuine concern but rather an undefined mixture.
•Distinct Purpose — A concern serves a specific purpose that differs from other concerns. While concerns may interact, each addresses a different fundamental question or responsibility.
•Potential for Isolation — At least in principle, a concern could be developed, tested, or modified without deep entanglement with other concerns (though full isolation isn't always practical).
•Stable Boundary — Over time, the boundary of a well-identified concern remains relatively stable, even as implementation details evolve.

Concerns Are Not Always Obvious

Categories of Concerns

Functional vs Non-Functional Concerns:

The most fundamental distinction is between concerns that define what the system does versus how well it does it:

Functional vs Non-Functional Concerns
Category	Definition	Examples	Typical Questions
Functional Concerns	What the system does—its features and capabilities	User registration, payment processing, order fulfillment, report generation	What actions can users perform? What data transformations occur?
Non-Functional Concerns	How well the system performs its functions—quality attributes	Performance, security, reliability, scalability, usability, maintainability	How fast? How secure? How available? How easy to modify?

Domain Concerns vs Technical Concerns:

Another critical distinction separates concerns related to the problem domain from concerns related to the technical solution:

Domain Concerns

•Business rules — How the domain works (e.g., 'orders over $100 get free shipping')
•Domain entities — Core concepts like Customer, Order, Product, Account
•Domain services — Operations meaningful in the business context (e.g., 'process refund')
•Workflows — Business processes that span multiple operations
•Policies — Business decisions encoded as logic (pricing, eligibility, etc.)

Technical Concerns

•Persistence — How data is stored and retrieved (databases, files)
•Communication — How components exchange data (HTTP, messaging)
•Infrastructure — Deployment, scaling, monitoring capabilities
•Presentation — How information is displayed to users (UI/UX)
•Integration — How external systems are connected and managed

The Volatility Principle

Core Application Concerns

The Universal Concerns:

Consider a typical web application processing a user request. Even in a simple CRUD operation, numerous concerns are involved:

concerns-in-action.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
// A seemingly simple function that actually addresses many concerns
async function createOrder(request: CreateOrderRequest): Promise<OrderResponse> {
    // CONCERN: Input Validation
    // Ensuring data meets structural and semantic requirements
    validateOrderRequest(request);
    
    // CONCERN: Authentication
    // Verifying the identity of the requester
    const user = await authenticateRequest(request.token);
    
    // CONCERN: Authorization
    // Determining if the authenticated user can perform this action
    await authorizeOrderCreation(user, request);
    
    // CONCERN: Business Logic / Domain Rules
    // Applying domain-specific calculations and validations
    const orderTotal = calculateOrderTotal(request.items);
    const discounts = applyDiscountRules(orderTotal, user);
    const finalTotal = orderTotal - discounts;
    
    // CONCERN: Inventory Management
    // Checking and reserving stock
    await reserveInventory(request.items);
    
    // CONCERN: Persistence
    // Storing data durably
    const order = await orderRepository.save({
        userId: user.id,
        items: request.items,
        total: finalTotal,
        status: 'pending'
    });
    
    // CONCERN: External Integration
    // Communicating with payment provider
    await paymentService.authorize(order.id, finalTotal);
    
    // CONCERN: Event Publishing / Notification
    // Informing other parts of the system
    await eventBus.publish('order.created', { orderId: order.id });
    
    // CONCERN: Logging / Observability
    // Recording what happened for debugging and analytics
    logger.info('Order created', { orderId: order.id, userId: user.id });
    
    // CONCERN: Response Formatting / Presentation
    // Preparing the output for the caller
    return formatOrderResponse(order);
}

The code above is instructive but problematic from a separation of concerns perspective. All these concerns are intertwined in a single function, making it difficult to:

Test any single concern in isolation
Modify one concern without risking others
Reuse individual concerns in other contexts
Understand what the function does at a glance

Recognizing these concerns is the first step. Later in this module, we'll learn techniques to isolate them.

Catalog of Common Concerns

•Input Handling — Parsing, deserializing, and validating incoming data before it enters the system
•Authentication — Verifying identity claims (who is making the request)
•Authorization — Enforcing access control (what are they allowed to do)
•Business Logic — Core domain rules and calculations that define the system's purpose
•Data Access — Reading from and writing to persistent storage
•Caching — Storing computed results to avoid redundant work
•Transaction Management — Ensuring atomicity and consistency of state changes
•Error Handling — Detecting, categorizing, and responding to failures
•Logging — Recording events for debugging, auditing, and analytics
•Metrics / Monitoring — Measuring system behavior for operational visibility
•Event Publishing — Notifying other components or systems of state changes
•Output Formatting — Transforming internal data into appropriate output formats

Techniques for Identifying Concerns

Identifying concerns is part analysis, part intuition—and it improves dramatically with practice. The following techniques provide systematic approaches to uncovering concerns in any system.

Technique 1: The 'Why' Chain

For any piece of code or functionality, ask 'Why does this exist?' repeatedly until you reach a fundamental purpose. Each distinct answer often reveals a separate concern.

Each level reveals a different aspect that might warrant separate consideration.

why-chain-analysis.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// Original mixed concern code
function processPayment(payment: Payment) {
    // Why? To ensure the amount is valid (VALIDATION)
    if (payment.amount <= 0) throw new Error("Invalid amount");
    
    // Why? To track what happened (LOGGING)
    console.log(`Processing payment: ${payment.id}`);
    
    // Why? To check user permission (AUTHORIZATION)
    if (!payment.user.canMakePayments) throw new Error("Not authorized");
    
    // Why? To persist the payment record (PERSISTENCE)
    database.save(payment);
    
    // Why? To communicate with payment provider (INTEGRATION)
    stripeClient.charge(payment.amount, payment.user.card);
    
    // Why? To inform other systems (EVENTING)
    eventBus.emit('payment.processed', payment);
}
 
// Each "Why?" answer points to a distinct concern that could be separated

Technique 2: The Stakeholder Lens

Different stakeholders care about different aspects of a system. Viewing the system through each stakeholder's perspective reveals concerns that might otherwise be invisible:

End Users see features, usability, performance
Business Owners see revenue, compliance, competitive advantage
Operations Teams see deployability, monitoring, incident response
Developers see maintainability, testability, architectural clarity
Security Teams see vulnerabilities, access control, audit trails
Data Teams see data quality, analytics, privacy compliance

Each stakeholder perspective illuminates different concerns.

The Change Driver Test

Technique 3: The Verbal Description Test

Describe what a component does in plain English. If you find yourself using the word 'and' frequently, you're likely describing multiple concerns:

Poor (multiple concerns): "This class handles user registration and sends welcome emails and logs the new user and validates email uniqueness."

Better (focused concerns):

"This class orchestrates the user registration process."
"This class sends templated emails to users."
"This class validates email uniqueness against the database."
"This class logs significant user events."

Each simple description points to a single concern.

Concern Identification Checklist

•Can you name this functionality in 2-3 words? If not, it may be multiple concerns.
•Would this functionality make sense in a different application? Universal applicability often indicates a separable concern.
•Does this functionality change for a single, identifiable reason? Multiple change drivers suggest multiple concerns.
•Can you imagine testing this aspect independently? Testability indicates separability.
•Would a different team member naturally own this? Organizational ownership often aligns with concerns.
•Does this functionality have its own vocabulary or domain terms? Distinct terminology suggests a distinct concern.

Granularity of Concerns

Concern Hierarchy:

Concerns can be analyzed at progressively finer levels of detail:

Levels of Concern Granularity
Level	Scope	Examples	Typical Separation Mechanism
System	Entire product or platform	E-commerce vs CMS vs Analytics	Separate systems, APIs, deployments
Subsystem/Bounded Context	Major functional area	Ordering, Inventory, Payments, Shipping	Microservices, modules, packages
Component/Service	Coherent functional unit	Order validation, price calculation, notification	Classes, services, functions
Aspect	Cross-cutting behavior	Logging, security, transaction management	Aspects, middleware, decorators
Method/Function	Single operation	Calculate discount, format date, validate email	Functions, private methods

granularity-examples.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
// SYSTEM LEVEL: Entire products/platforms
// - OrderingSystem
// - InventorySystem
// - PaymentGateway
 
// SUBSYSTEM LEVEL: Bounded contexts within a product
namespace OrderingContext {
    // Everything related to orders
    class OrderService { /* ... */ }
    class OrderRepository { /* ... */ }
    class OrderEvents { /* ... */ }
}
 
namespace PaymentContext {
    // Everything related to payments  
    class PaymentService { /* ... */ }
    class PaymentRepository { /* ... */ }
    class PaymentEvents { /* ... */ }
}
 
// COMPONENT LEVEL: Services within a bounded context
class OrderService {
    constructor(
        private validator: OrderValidator,      // Validation concern
        private calculator: PriceCalculator,    // Calculation concern
        private repository: OrderRepository,    // Persistence concern
        private events: OrderEventPublisher     // Event concern
    ) {}
}
 
// ASPECT LEVEL: Cross-cutting concerns
// Applied via decorators/middleware
@Transactional()
@Logged()
@Cached({ ttl: 300 })
async getOrderById(id: string): Promise<Order> { ... }
 
// METHOD LEVEL: Single operations
class PriceCalculator {
    calculateBasePrice(items: Item[]): Money { /* focused logic */ }
    applyDiscounts(price: Money, discounts: Discount[]): Money { /* focused logic */ }
    applyTax(price: Money, region: Region): Money { /* focused logic */ }
}

Finding the Right Granularity

Concerns Across Programming Paradigms

The concept of concerns transcends any specific programming paradigm, but different paradigms offer different mechanisms for expressing and separating concerns.

Object-Oriented Programming:

OOP provides classes and interfaces as the primary mechanism for separating concerns. Each class ideally represents a single concern, and interfaces define the contracts between concerns:

oop-concerns.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
// OOP: Classes encapsulate concerns
interface OrderRepository {
    save(order: Order): Promise<void>;      // Persistence concern
    findById(id: string): Promise<Order>;
}
 
interface PaymentProcessor {
    process(amount: Money): Promise<PaymentResult>;  // Payment concern
}
 
interface NotificationService {
    notify(user: User, message: Message): Promise<void>;  // Notification concern
}
 
// Concerns are composed via dependency injection
class OrderService {
    constructor(
        private repository: OrderRepository,
        private payment: PaymentProcessor,
        private notifications: NotificationService
    ) {}
    
    async createOrder(order: Order): Promise<void> {
        // This method orchestrates but doesn't implement the concerns
        await this.repository.save(order);
        await this.payment.process(order.total);
        await this.notifications.notify(order.user, orderConfirmation(order));
    }
}

Functional Programming:

FP separates concerns through pure functions and composition. Each function addresses a single transformation, and concerns are separated by avoiding side effects:

fp-concerns.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
// FP: Functions represent concerns as pure transformations
// Each function handles one concern
 
// Validation concern (pure)
const validateOrder = (order: Order): Either<ValidationError[], Order> => {
    const errors = [
        validateItems(order.items),
        validateCustomer(order.customer),
        validateShipping(order.shipping)
    ].filter(isError);
    
    return errors.length > 0 ? left(errors) : right(order);
};
 
// Calculation concern (pure)  
const calculateTotal = (order: Order): OrderWithTotal => ({
    ...order,
    total: order.items.reduce((sum, item) => sum + item.price * item.quantity, 0)
});
 
// Discount concern (pure)
const applyDiscounts = (order: OrderWithTotal): OrderWithTotal => ({
    ...order,
    total: order.total * (1 - getDiscountRate(order))
});
 
// Composition: pipeline of concerns
const processOrder = pipe(
    validateOrder,
    map(calculateTotal),
    map(applyDiscounts),
    chain(persistOrder),    // Side effect isolated
    chain(sendNotification)  // Side effect isolated
);

Aspect-Oriented Programming:

AOP provides explicit mechanisms for separating cross-cutting concerns (those that span multiple components) from primary logic:

aop-concerns.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
// AOP: Cross-cutting concerns declared separately from business logic
 
// Business logic - clean and focused
class OrderService {
    async createOrder(order: Order): Promise<Order> {
        const validated = this.validator.validate(order);
        const priced = this.calculator.calculateTotal(validated);
        return this.repository.save(priced);
    }
}
 
// Cross-cutting concerns declared as aspects
@Aspect()
class LoggingAspect {
    @Before('execution(* OrderService.*(..))')
    logEntry(jp: JoinPoint) {
        console.log(`Entering ${jp.methodName} with args: ${jp.args}`);
    }
    
    @After('execution(* OrderService.*(..))')
    logExit(jp: JoinPoint) {
        console.log(`Exiting ${jp.methodName}`);
    }
}
 
@Aspect()
class TransactionAspect {
    @Around('execution(* *Repository.save(..))')
    wrapInTransaction(jp: JoinPoint) {
        const tx = this.transactionManager.begin();
        try {
            const result = jp.proceed();
            tx.commit();
            return result;
        } catch (e) {
            tx.rollback();
            throw e;
        }
    }
}

Paradigm Agnostic Principle

Identifying Concerns: Practical Exercises

Let's apply what we've learned to analyze real code and identify its concerns. These exercises build the analytical muscle needed for effective separation.

Exercise 1: The God Function

Examine the following function and identify all distinct concerns:

exercise-god-function.ts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
async function handleUserRegistration(req: Request, res: Response) {
    // What concerns can you identify here?
    const startTime = Date.now();
    
    const { email, password, name } = req.body;
    
    if (!email || !email.includes('@')) {
        return res.status(400).json({ error: 'Invalid email' });
    }
    
    if (!password || password.length < 8) {
        return res.status(400).json({ error: 'Password too short' });
    }
    
    try {
        const existing = await db.users.findOne({ email });
        if (existing) {
            return res.status(409).json({ error: 'Email already registered' });
        }
        
        const hashedPassword = await bcrypt.hash(password, 10);
        
        const user = await db.users.insert({
            email,
            password: hashedPassword,
            name,
            createdAt: new Date()
        });
        
        const token = jwt.sign({ userId: user.id }, SECRET, { expiresIn: '24h' });
        
        await sendEmail({
            to: email,
            subject: 'Welcome!',
            body: `Welcome to our platform, ${name}!`
        });
        
        await analytics.track('user.registered', { userId: user.id });
        
        console.log(`User ${email} registered in ${Date.now() - startTime}ms`);
        
        res.cookie('authToken', token, { httpOnly: true });
        res.status(201).json({ user: { id: user.id, email, name } });
        
    } catch (error) {
        console.error('Registration failed:', error);
        await alerting.notify('registration-failure', error);
        res.status(500).json({ error: 'Registration failed' });
    }
}

Concerns in the Registration Function

•Request Parsing — Extracting data from the HTTP request body
•Input Validation — Checking email format and password length
•Uniqueness Verification — Checking if email is already registered
•Password Security — Hashing passwords before storage
•User Persistence — Storing the user in the database
•Token Generation — Creating an authentication JWT
•Email Notification — Sending the welcome email
•Analytics Tracking — Recording the registration event
•Performance Logging — Timing and logging the operation
•Response Formatting — Setting cookies and JSON response
•Error Handling — Catching and responding to failures
•Alerting — Notifying operations of failures

The Separation Challenge

Summary: Identifying Concerns

Identifying concerns is the foundational skill for separation of concerns. Before you can separate, you must see. Let's consolidate what we've learned:

Key Takeaways

•A concern is a coherent area of interest — any distinct aspect of functionality that can be identified, named, and discussed independently.
•Concerns exist in categories — functional vs non-functional, domain vs technical. Each category brings different considerations.
•Common concerns recur across systems — authentication, validation, persistence, logging, and others appear in nearly every application.
•Use systematic techniques — the 'Why' chain, stakeholder lens, change driver test, and verbal description test help uncover hidden concerns.
•Granularity matters — concerns exist at system, subsystem, component, aspect, and method levels. Choose appropriate granularity for separation.
•Programming paradigms offer different tools — OOP uses classes, FP uses pure functions, AOP uses aspects. The principle transcends the mechanism.

What's Next:

Page Complete

1 / 4