Link to video

A cache is a hardware or software component that stores data so that future requests for that data can be served faster; the data stored in a cache might be the result of an earlier computation or a copy of data stored elsewhere.

Reasons why you need cache :

1. Reduce Network Calls.
2. Avoid recomputations.
3. Avoid load on DB

Eviction Policies for Cache :

• LRU
• LFU
• Sliding Window Based Policy

Problems with caching :

Thrashing is constantly inputting and outputting into the cache without using the results. This is expensive as well as harmful. Another problem is data consistency.

Where do you place the cache ? Closer to the database or the server ?

If you place close to server, place it in-memory. But if you do this if server fails, the in-memory cache also fails. It is for sure faster but if the information is financial data, it is not a good method.

Another way is to have distributed global caches, benefits are faster and avoid DB load. Something like redis(persistent storage which is like a cache). More accurate than in-memory cache. More resilient and scale independently as well.

Write through or Write Back cache

1. Write Through
   • Write on cache directly before writing on database.
   • It will cause issues if same cache is used by multiple servers, suppose you update cache for server 1 but not for other. Data consistency will be lost.
2. Write Back
   • Hit database directly then make sure you make an entry in cache.
   • Cache invalidation will be expensive.