Skip to Main Content
With Cloud Computing, scaling compute-intensive applications out to thousands of nodes has become a matter of minutes. If with an increasing number of users the system throughput reaches inacceptable values, additional computational power can be purchased on demand and "as you go". Thereby, the newly added instance nodes channel user requests to yet more servers and this way distribute the entire system load across an increased number of physical machines. Even though the standard cloud approach aims at replicating business logic to scale out, we believe that leveraging cloud caches can result into even better and yet cheaper scaling. While only one cache instance can often replace multiple service instances, the charged prices for either instance are the same. Throughout this paper, we are going to examine how by using cloud caches system responsiveness can be greatly improved. Furthermore, we are going to show how an at least similar level of throughput can be achieved at a significantly reduced system TCO, by introducing a distributed caching architecture to an existing cloud application.