Read-Through vs. Write-Through (Cache)

Question 1

Why would you care about the implementation of a read-through vs a write-through cache?

Answer 1

Choosing between read-vs-write can be an important optimization where speed is critical.

Question 2

How does a read-through cache work? Can you walk through the process?

Answer 2

A read-through cache is populated on-demand.

Basically when a request is made, the cache would:
- Check for the requested item
- If missed, it would query primary storage, store in the cache, and serve
- If hit, it would serve directly

Question 3

What are the pros of a read-through cache?

Answer 3

It’s transparent (since the application only interacts with the cache), highly performant on hits (served directly from the cache), and reduces primary storage pressure (frequent reads don’t hit primary storage).

Question 4

What’s the single con of a read-through cache?

Answer 4

Misses result in a latency hit as the data must be read from primary storage.

Question 5

How does a write-through cache work? Can you walk through the process?

Answer 5

In a write-through cache, all write operations are written to the cache AND primary storage.

When a write request is made:
- Data is written to the cache
- Data is written to the primary storage
- All reads are made from the cache

Question 6

What are the three primary pros of using write-through cache?

Answer 6

A write-through cache provides:
- Consistency guarantees as the cache and primary storage are always in sync
- No data is lost during crashes as both stores have the same data
- Simplfied writes as it’s written to the cache and primary storage

Question 7

What are the two cons of a write-through cache?

Answer 7

Writes can have higher latency as they must be written to two places instead of one, and since every write hits primary storage it can increase pressure on it.

Question 8

What type of systems would favor a read-through cache?

Answer 8

Most read-heavy systems would prefer a read-through, especially if data isn’t frequently updated.

Question 9

What systems would prefer a write-through cache?

Answer 9

Systems that favor consistency or data integrity would favor a write-through cache, even at the cost of increased write latency.

Question 10

What are two good reasons to use a read-through cache?

Answer 10

Ready-heavy with infrequent writes and eventually consistent with reads.

Question 11

What are two good reason to use a write-through cache?

Answer 11

Systems where data consistency is paramount and updates must be immediately propagated.