[Cache Design] Is the most basic multi-level cache suitable and how should it be set up?

Recently, we want to build a level-one caching mechanism for our system. But I always feel like something is missing.

Environment:
Load balancing, master-slave separation, Redis standalone (can be multiple machines in the future)

Now the initial idea:

<code>浏览器缓存-》本地文件缓存-》内存缓存（Redis）-》Db
</code>

After the user accesses the web application, set the browser cache for it, and then set the local file cache and memory cache.
After other users visit, I think the steps are as follows:

1. Check if there is browser cache

2. Retrieve whether the local machine has a file cache

3. Memory cache

4. Db

My question is:

But I feel that something is missing in a certain step, or I feel that the (multi-level) cache expiration time is difficult to choose.

Also, is the local file cache {checking expiration time, reading (deleting, generating) files} worth it compared to a single connection that jumps directly to the memory cache (Redis)?

So I would like to consult my basic caching mechanism to see if it is suitable or if there are any shortcomings that can be improved. Thank you!

Reply content:

Recently, we want to build a level-one caching mechanism for our system. But I always feel like something is missing.

Environment:
Load balancing, master-slave separation, Redis standalone (can be multiple machines in the future)

Now the initial idea:

<code>浏览器缓存-》本地文件缓存-》内存缓存（Redis）-》Db
</code>

After the user accesses the web application, set the browser cache for it, and then set the local file cache and memory cache.
After other users visit, I think the steps are as follows:

1. Check if there is browser cache

2. Retrieve whether the local machine has a file cache

3. Memory cache

4. Db

My question is:

But I feel that there is something missing in a certain step, or I feel that the (multi-level) cache expiration time is difficult to choose.

Also, is the local file cache {checking expiration time, reading (deleting, generating) files} worth it compared to a single connection that jumps directly to the memory cache (Redis)?

So I would like to consult my basic caching mechanism to see if it is suitable or if there are any shortcomings that can be improved. Thank you!

1. Multi-level cache can reduce the pressure on the system and greatly reduce RT. However, one aspect that needs to be considered is the management of multi-level cache. This is also mentioned by the author in the article. This is avoided by using multi-level cache. Can't solve the problem. As for how to invalidate multi-level cache, you can try to use a local timer to refresh the cache at intervals

2. In fact, the file cache can be replaced by a local memory cache. It is also possible to design it as a file cache. However, when the volume of local disk I/O is large, I am afraid that it cannot handle it. As for which one is more efficient and network overhead, it needs to be based on the actual situation. Go and take a pressure test

3. Multi-level caching is more about cache penetration and program robustness. When there is a problem with the centralized cache, our application can continue to run; some hot data is made into a memory cache, so there is no need to access the centralized cache. , which can reduce the pressure on centralized cache. So in this aspect, file caching is better than Redis’s centralized caching