发现一篇关于cache strategies的好文

发现这篇关于cache strategies文章讲的很好
https://codeahoy.com/2017/08/11/ ... oose-the-right-one/
我把要点记录下来了
Cache Aside
  Application checks from cache, if hit, returns data. If miss, reads from DB, returns data to client and updates cache.

  Use Case
    read-heavy workload

  Pros
    resilient to cache failures
    data model in cache can be different from DB

  Cons
    If write strategy is to write to DB directly, cache might be not consistent with DB (using TTL)
    If consistency is required, either invalidate cache entry, or use write through.

Read Through
  When cache miss, cache provider/library loads data from DB, populates cache and return data

  Diff from Cache Aside
    loading data is done by library or stand-alone cache provider
    data model has to be the same as in DB

  Use Case
    read-heavy workload

  Cons
    cache miss for every first-time request (can do warming or pre-heating)
    If write strategy is to write to DB directly, cache might be not consistent with DB (using TTL)

Write Through
  Write to cache and then cache provider/library writes data to DB

  Use Case
    combining Read Through and Write Through gets all the benefits of read-through and we also get data consistency guarantee, freeing us from using cache invalidation techniques. (DynamoDB Accelerator uses this)

Write Around
  Data is written to DB directly

  Use Case
    Write-around can be combine with read-through and provides good performance in situations where data is written once and read less frequently or never. For example, real-time logs or chatroom messages. Likewise, this pattern can be combined with cache-aside as well.

Write Back (Write Behind)
  Application writes data to cache and immediatly ack, after some time, cache write data back to DB

  Use Case
    Good for write-heavy workloads
    Resilient to DB failure
    If batching or coalescing is supported, reduce overall write to DB
    Redis use cache-aside and write-back to better absort spikes during peak load
    Most relational DB storage engines (i.e. InnoDB) have write-back enabled by default

  Cons
    If cache failure, data maybe permanently lost

求米～