Monitoring Latency Sensitive Enterprise Applications on the Cloud

Enterprises are increasingly deploying their applications in the cloud given the cost-saving advantages and the potential to geo-distribute applications to ensure resilience and better service experience. Most of the enterprise applications have stringent latency constraints for better user experience or by business requirements. One of the fundamental challenges in the deployment of such geo-distributed enterprise applications is to monitor their performance once deployed. On the one hand, it is clear that user observed latency is the most important parameter and enterprises should be able to efficiently measure the end-to-end latency at the granularity of every user transaction, on the other hand, they should also be able to isolate the problem to each component to optimize cost and improve utilization. It is therefore crucial for any monitoring framework to provide latency measurements at both the levels of granularity. Also the monitoring framework should be easy to integrate with any application and have minimal impact on the application in order to be scalable. In this work, we present the design and implementation of such a monitoring framework for latency sensitive applications on the cloud. We also evaluate our framework using benchmark applications deployed on two prominent cloud service providers, Amazon AWS and Microsoft Azure.