Skip to content
advanced Phase 85 · Distributed Fundamentals

Distributed Caching

Distributed caching with Redis cluster, cache consistency, and cache warming strategies

45m
0 problems
Topic Progress 0%

Redis Cluster Caching

Cluster Architecture

┌─────────────┐  ┌─────────────┐  ┌─────────────┐
│  Node 1     │  │  Node 2     │  │  Node 3     │
│  Slots 0-   │  │  Slots 5461-│  │  Slots 10923│
│  5460       │  │  10922      │  │  16383      │
└──────┬──────┘  └──────┬──────┘  └──────┬──────┘
       │                │                │
┌──────┴──────┐  ┌──────┴──────┐  ┌──────┴──────┐
│  Replica 1  │  │  Replica 2  │  │  Replica 3  │
└─────────────┘  └─────────────┘  └─────────────┘

Cache Distribution

Key Hash: CRC16(key) MOD 16384 = slot number
Each node responsible for subset of slots
Replicas provide redundancy per node

Magento Cache Config

// app/etc/env.php
'cache' => [
    'frontend' => [
        'default' => [
            'backend' => 'Magento\Framework\Cache\Backend\Redis',
            'backend_options' => [
                'server' => 'redis-cluster.example.com',
                'port' => '6379',
                'database' => '0',
                'compress_data' => '1'
            ]
        ],
        'page_cache' => [
            'backend' => 'Magento\Framework\Cache\Backend\Redis',
            'backend_options' => [
                'server' => 'redis-cluster.example.com',
                'port' => '6379',
                'database' => '1',
                'compress_data' => '0'
            ]
        ]
    ]
],

Cache Consistency

Consistency Challenges

Problem: Cache-DB inconsistency
1. Write to DB
2. Invalidate cache
3. Read request gets stale data from cache
4. Cache miss → read from DB → inconsistent

Consistency Strategies

Strategy           | Approach                | Trade-off
───────────────────|─────────────────────────|─────────────
Write-through      | Update cache + DB       | Write latency
Write-behind       | Update cache, async DB  | Possible loss
Cache-aside        | Read from DB, cache     | Stale reads
Event-driven       | Invalidate on events    | Eventual consistency

Event-Driven Invalidation

// Invalidate cache on data change
$eventManager->dispatch('catalog_product_save_after', [
    'product' => $product
]);

// Observer invalidates cache
public function execute(EventObserver $observer) {
    $product = $observer->getEvent()->getProduct();
    $cacheKey = 'product_' . $product->getId();
    $this->cache->remove($cacheKey);
    
    // Also invalidate related caches
    $this->cache->clean([
        'catalog_product_' . $product->getId(),
        'category_products_' . $product->getCategoryId()
    ]);
}

Cache Warming

Warming Strategies

Strategy          | When                  | Use Case
──────────────────|───────────────────────|─────────────
Pre-deployment     | Before traffic        | Code deploy
Scheduled          | Off-peak hours        | Daily warm
On-demand          | On cache miss spike   | Flash sales
Predictive         | Based on patterns     | Smart warming

Cache Warming Script

// Warm critical caches
class CacheWarmer {
    public function warm() {
        // Product pages
        $products = $this->productCollection->create()
            ->addFieldToFilter('status', 1)
            ->setPageSize(1000)
            ->load();
        
        foreach ($products as $product) {
            $this->warmProductPage($product->getId());
        }
        
        // Category pages
        $categories = $this->categoryCollection->create()
            ->addFieldToFilter('is_active', 1)
            ->load();
        
        foreach ($categories as $category) {
            $this->warmCategoryPage($category->getId());
        }
    }
    
    private function warmProductPage($productId) {
        $url = $this->urlBuilder->getUrl('catalog/product/view', [
            'id' => $productId
        ]);
        $this->httpClient->get($url);
    }
}

Scheduled Warming

# cron: Warm cache daily at 3 AM
0 3 * * * /usr/local/bin/magento-cache-warm.sh

# Warm before deployment
/bin/magento cache:warm

Cache Failure Handling

Cache Failure Scenarios

Scenario           | Impact       | Handling
───────────────────|──────────────|──────────────────
Cache miss         | DB load      | Normal operation
Cache unavailable  | All DB       | Graceful degradation
Cache stampede     | DB overload  | Prevent with locks
Cache inconsistency| Stale data   | Event-driven invalidation

Cache-Aside with Fallback

function getFromCache($key, $ttl = 3600) {
    try {
        $value = $this->redis->get($key);
        if ($value !== false) {
            return unserialize($value);
        }
    } catch (Exception $e) {
        $this->logger->warning('Cache read failed', ['key' => $key]);
    }
    
    // Cache miss or failure - get from DB
    $value = $this->database->fetch($key);
    
    try {
        $this->redis->setex($key, $ttl, serialize($value));
    } catch (Exception $e) {
        $this->logger->warning('Cache write failed', ['key' => $key]);
    }
    
    return $value;
}

Cache Stampede Prevention

// Use locks to prevent stampede
$lockKey = 'lock_' . $cacheKey;
if ($this->redis->set($lockKey, '1', ['NX', 'EX' => 10])) {
    // Got lock - rebuild cache
    $value = $this->rebuildCache($cacheKey);
    $this->redis->del($lockKey);
} else {
    // Wait for other process
    usleep(100000); // 100ms
    return $this->getFromCache($cacheKey);
}

Quiz

1. How does Redis Cluster distribute keys?

Question 1 options

2. What is cache stampede?

Question 2 options

3. What is cache warming?

Question 3 options

Flashcards

Question

Redis Cluster key distribution?

Click to reveal answer

Answer

CRC16(key) MOD 16384 assigns to slot

Question

Cache stampede?

Click to reveal answer

Answer

Multiple processes rebuilding same cache simultaneously

Question

Cache warming purpose?

Click to reveal answer

Answer

Pre-populate cache before traffic to avoid cold starts

Question

Cache consistency strategy?

Click to reveal answer

Answer

Event-driven invalidation on data changes

Revision Notes

Key Takeaways

  • 1. Redis Cluster distributes keys across 16384 hash slots
  • 2. Event-driven invalidation ensures cache consistency
  • 3. Cache warming pre-populates before traffic hits
  • 4. Use locks to prevent cache stampede
  • 5. Graceful degradation when cache fails

Interview Tips

  • • Explain Redis Cluster key distribution
  • • Discuss cache consistency strategies and trade-offs
  • • Describe cache warming and stampede prevention

Cheat Sheet

Distributed Caching:
  Redis Cluster: 16384 hash slots
  Key → Slot → Node assignment

Consistency:
  Write-through: Sync cache + DB
  Event-driven: Invalidate on changes
  Cache-aside: Read-through pattern

Cache Warming:
  Pre-populate before traffic
  Scheduled: Daily at off-peak
  On-demand: Before deployments

Failures:
  Cache-aside with DB fallback
  Locks prevent stampede
  Graceful degradation