• Why do we need caching?
• When to use caching?
• When not to use caching?
• How to implement it?
• Cache writing policies

Why do we need caching?

• To speed up the system. Avoid redoing the same computationally complex operations. To reduce the latency.
• Caching does this by storing a “cached” copy of the data, in a smaller, faster data store.
• When data is immutable caching works great, but when data is mutable we need to update data in DB and in cache.
• Cache eviction policy — specify time that the data will be cached. Refresh rate — how to update the data in cache.

When to use caching?

• Slow to access
• A lot of network requests
• Computationally long operations
• Slow hardware
• Static or slow to change data

When not to use caching?

• Time to access cache = time to access real data
• When low repetition, because caching performance comes from repeated memory access patterns.
• Data changes frequently

How to implement it?

• Client level caching, no need to go to server. Example: questions list on AlgoExpert.
• Server level caching, no need to go to database
• Middleware

Cache writing policies