<h1>Caching</h1>
<p>You all probably already know what caching is. It is essentially a store from where you can retrieve your data faster than a primary source. Caching is one of the most important UX optimization techniques, if used correctly.</p>
<p>Let’s look at some common caching architectures.</p>
<hr>
<h1>Caching Architectures</h1>
<h2>Read Caching Architectures</h2>
<h3>Cache Aside</h3>
<p>Return if the data is present in cache, otherwise retrieve from database and update the cache.</p>
<h3>Read Ahead</h3>
<p>The cache reads from the database ahead of the user’s request.</p>
<hr>
<h2>Write Caching Architectures</h2>
<h3>Write Through</h3>
<p>Data is written to cache first and cache synchronously writes data to DB. Suffers from the risk of inconsistency.</p>
<h3>Write Behind</h3>
<p>Similar to Write Through but asynchronous write to DB.</p>
<hr>
<h1>Cache Eviction Strategies</h1>
<h2>Least Frequently Used (LFU)</h2>
<p>Evicts the least frequently accessed data.</p>
<h2>Least Recently Used (LRU)</h2>
<p>Evicts the least recently accessed data.</p>
<h2>First-In-First-Out (FIFO)</h2>
<p>Evicts the oldest inserted data first.</p>
<h2>TTL (Time-To-Live)</h2>
<p>Timer-based expiry for cached data.</p>
<hr>
<h1>Common Problems</h1>
<h2>Thundering Herd Problem / Cache Stampede</h2>
<p>Suppose a key is about to expire while there are plenty of requests coming in which were handled by this cache. Now when the key becomes invalid, all the requests will overwhelmingly hit the database, causing cascading failures.</p>
<h3>Solutions</h3>
<h4>1. Request Coalescing</h4>
<p>If all requests require the same cache key, allow only one request to go through while the other requests wait to be served from the cache.</p>
<h4>2. Prefetching</h4>
<p>Prefetch to update the key and do not let it expire if it is a popular key.</p>
<hr>
<h2>Cache Consistency</h2>
<p>Depending upon the consistency requirements, we can either:</p>
<h3>Let the data become eventually consistent</h3>
<p>or</p>
<h3>Invalidate cache on write</h3>
<hr>
<h1>Conclusion</h1>
<p>Cache is absolutely essential for performance, but can be more harmful than good if used without taking into consideration the pros and cons.</p>
<p>Thanks for reading!</p>
<p>#redis #memcached #system-design #beginner #performance #API</p>


Lakshya Mehta

A beginner-friendly overview of caching fundamentals covering read and write architectures (Cache Aside, Read Ahead, Write Through, Write Behind), cache eviction strategies (LFU, LRU, FIFO, TTL), and common problems like the Thundering Herd/Cache Stampede with solutions such as request coalescing and prefetching. Also touches on cache consistency approaches.

Caching - A Beginner's Breakdown