Data ingestion is fast and data life becomes very short. When a user searches a product in a retailer website, deriving recommendations at that point of time makes sense to the business to cross-sell other related products. The customer pulse is need to be understood, and IoT is a key driver to understand the market. Online retailers are seeking a solution for sudden data surge scalability protection during special events (planned /unplanned) like festive seasons to support a sudden increase in concurrent user base. With the large concurrent user base, Microservice based solution and application deployment are gaining an edge and soon it will become a defacto standard. With the increase of cloud usage (Platform, User, Service, etc.), industries are looking at solutions for more and more low cost, high scalable application deployment platform (PaaS) to make it a win-win deal with cloud providers. In this paper, the authors have proposed an agent-based architecture for auto scaling of an application deployed on the premises or in PaaS based on data surge sense at minimal cost without compromising on performance and other non-functional requirements in real time. To address this research issue, microservice based Twitter sentiment analysis application is created to ingest realtime data. Then this approach will define the guidelines to auto scale (Lorido-Botran et al., 2014) up/down the application based on incoming data surge to define dynamic and adaptive resource utilization, which in turn reduces infrastructure and service utilization cost of the cloud-based platform.