What is Caching? - Your Speed Multiplier

🎯 Challenge 1: The Library Puzzle

Scenario: You're a librarian. A student asks for "Harry Potter and the Sorcerer's Stone" - a book requested 20 times per day.

Method A: The Thorough Approach

• Walk to back storage room (5 minutes)
• Search through shelves (3 minutes)
• Retrieve book (2 minutes)
• Walk back to front desk (5 minutes)
• Total: 15 minutes per request
• Daily time: 15 min × 20 requests = 300 minutes (5 hours!)

Method B: The Smart Approach

• Keep a copy of popular books at the front desk
• Student asks → grab from desk (10 seconds!)
• Total: 10 seconds per request
• Daily time: 10 sec × 20 requests = 200 seconds (3 minutes!)

Question: Which method serves students better? How much time do you save?

Answer: Method B saves 297 minutes (almost 5 hours!) daily - that's 99.97% faster!

The Answer: This is Caching!

Definition: Caching stores frequently accessed data in a fast, nearby location so you don't have to retrieve it from the slow, distant source every time.

---

🧠 The Memory Hierarchy: Why Caching Exists

Your computer has different storage speeds:

**Speed comparison in human terms (**all theoretical comparisons only)

If accessing RAM takes 1 second: ()

• L1 Cache: 0.02 seconds (blink of an eye)
• SSD: 16 minutes (theoretically)
• HDD: 1.5 days
• Network: 1+ weeks

Key insight: Caching moves data from slow storage to fast storage!

---

🏪 Real-World Caching Analogy: The Grocery Store

Without Cache (Always go to the warehouse):

You need milk:

1. You drive to warehouse (30 min) 🚗
2. You search for milk (10 min) 🔍
3. You drive back home (30 min) 🚗 Total: 70 minutes

You need bread:

1. You drive to warehouse again (30 min) 🚗
2. You search for bread (10 min) 🔍
3. You drive back home (30 min) 🚗 Total: 70 minutes

Daily time for 5 items: 350 minutes! 😰

With Cache (Local grocery store):

You need milk:

1. You walk to corner store (2 min) 🚶
2. You grab the milk (1 min) 🥛
3. You walk back home (2 min) 🚶 Total: 5 minutes

You need bread:

1. You walk to corner store (2 min) 🚶
2. You grab the bread (1 min) 🍞
3. You walk back home (2 min) 🚶 Total: 5 minutes

Daily time for 5 items: 25 minutes! ⚡

Mental model: The corner store is your cache - it stocks popular items locally for quick access. The warehouse is your database - it has everything but it's far away!

---

💻 Caching in Web Applications

---

🎪 Types of Caching: Where Data Gets Stored

1. Browser Cache

Location: It is located in our computer Stores: It basically stores images, CSS, JavaScript, HTML Speed: Instant (already on your device) Example: Facebook logo doesn't re-download every page load

2. CDN Cache

Location: It is located at edge servers worldwide across different locations around the world Stores: It generally stores static content (images, videos, files) Speed: It is very fast (geographically close) Example: Netflix videos served from nearby servers

3. Server Cache

Location: It is a web server memory (Redis, Memcached) Stores: It generally stores database query results, session data, computed values Speed: Fast (in-memory) Example: User profile data, product listings

4. Database Cache

Location: It is located at the database memory Stores: It queries results, frequently accessed rows Speed: Fast (avoids disk I/O) Example: Most-read blog posts

5. Application Cache

Location: Application memory Stores: Object instances, computed results Speed: Very fast (in-process) Example: Configuration settings, lookup tables

---

🎮 Interactive Exercise: What Should You Cache?

Decide if each scenario should use caching:

Think about each one...

---

Answers:

1. Account Balance: ❌ NO (or very short TTL (time to live))

   • Account Balances changes frequently
   • It must always be accurate
   • Cache for 1-2 seconds max if needed
   • Mental model: Money must be precise!

2. Product Catalog: ✅ YES

   • A product catalog changes infrequently
   • It is read far more than written
   • We can cache it for 5-60 minutes
   • Mental model: Products don't change every second

3. Weather Forecast: ✅ YES

   • Weather forecast updates every 30-60 minutes
   • It is usually the same data for all users in city
   • Cache for 30 minutes
   • Mental model: Weather doesn't change instantly

4. Live Stock Prices: ❌ NO (or 1-second cache)

   • Stock prices changes every millisecond
   • Users expect real-time prices to carry out high frequency trade
   • Maybe 1-second cache for high traffic
   • Mental model: "Live" means fresh!

5. Profile Picture: ✅ YES

   • Rarely changes
   • Same for all viewers
   • Cache for 24 hours or longer
   • Mental model: People don't change photos often

---

📋 The Caching Decision Framework

Cache when data is:

• ✅ Read frequently (high traffic)
• ✅ Expensive to generate (complex queries)
• ✅ Relatively stable (doesn't change often)
• ✅ Identical for multiple users
• ✅ Acceptable if slightly stale

Don't cache when data is:

• ❌ Written more than read
• ❌ Must be 100% fresh (financial transactions)
• ❌ User-specific and rarely reused
• ❌ Cheap to generate
• ❌ Privacy-sensitive

The Golden Rule:

Cache value = (Access frequency × Generation cost) / Freshness requirement

If Cache value > Threshold → Cache it!

---

💡 Simple Caching Example

Without Cache: // Every request hits database

🚨 Common Caching Mistakes

| ❌ Mistake 1: Caching Everything

// BAD: Cache user's shopping cart

// Changes every second

// BAD: Cache current time

// Always stale

// Mental model: Don't cache what changes constantly !❌ Mistake

2: Never Expiring Cache

// BAD: Cache forever

// What if product prices change?

// GOOD: Set expiration (TTL - Time To Live)

Mistake 3: Cache Stampede

PROBLEM: Cache expires, 1000 requests hit database simultaneously

// All 1000 requests see cache miss

All 1000 query database at once

// Database overload! 💥

// SOLUTION: Use locking or cache warming

---

Common Mistakes

---

Key Takeaways

1. Caching stores frequently accessed data in fast, nearby storage — reducing latency from seconds to milliseconds
2. Cache when data is read-heavy, expensive to generate, and tolerant of slight staleness — don't cache rapidly changing or security-sensitive data
3. Multiple cache layers exist — browser cache, CDN cache, server cache (Redis/Memcached), database cache, and application cache
4. Cache stampede is a critical failure mode — when cache expires and thousands of requests hit the database simultaneously
5. Always set a TTL (Time To Live) — caching forever leads to stale data and memory exhaustion