High availability and scalable application clustering solution for a large-scale OLTP application
Mohit Nanda, Amol Khanapurkar, Prabin Ranjan Sahoo · 2011
The world today expects lightening-speed response times and 100% availability from an application. To sustain business and meet these demands the applications needs cost-efficient, scalable and reliable architectures. In spite of multiple commercial and open-source solutions being available for scaling-out; validating their claims and verifying their suitability for one's application is a non-trivial problem. There exists no proof outside vendor-benchmarks, verifying these claims. The vendor claims cannot be trusted verbatim, without careful evaluation of the same. Technical evaluation of any such solution is vital before its widespread adoption. It involves careful and meticulous design of test-cases which directly fit your application' requirements and cover all possible scenarios. In this paper, the authors have discussed, in detail, the methodology to do such an evaluation. After careful study of the alternatives, one such solution for application clustering and distributed caching was deployed on a large scale OLTP application. The solution was then evaluated by performing meticulous test-cases along with comparative evaluation with a non-clustered environment. The paper concludes with detailed analysis of the results, which validate the benefits of the solution. It further discusses the applicability of the results and solution, in larger context.