The write-through model assumes we are starting with an empty database and we are performing writes to the database, after which reads will start very shortly. We write each piece of data to the caching cluster and to the database, thus every read will potentially result in a cache hit and the read performance will be dramatically increased. This model is perfect for offloading databases with very simple queries, since performing complex transaction requests will still generate responses that cannot be cached through just a write process. Also, since we are caching all writes, this can be very heavy on the caching system and can cause a lot of cache churn and a lot of data that is never read to be written into the cache.