Budget optimisation without a quality loss


A customer in the e-commerce sector has been challenged in the formation of a stock goods listing based on various logistic parameters. An internal software development project of new stock goods listing sorting software has been arranged within the company. Unfortunately, after 40% of the planned development period, it has become evident that there is a very strong CPU shortage.  As a result, it is necessary to extend a server cluster in a customer’s private datacenter.  However, this extension has not been  completed by the operational support department (OPS) within the necessary period of time.


It was decided to move the development workload into the public computational cloud infrastructure (AWS). Unfortunately, there was neither enough experience with a migration of the necessary infrastructure to the cloud nor the necessary skillsets of the infrastructure management automation team for tasks such as semi-automatic cluster extension with modern devOps tools. An additional issue consisted of the incomplete understanding of the billing rules of the chosen cloud provider.


Since the computational volume was supposed to continue growing, a budget for cloud infrastructure has been planned using typical economic tow/low thresholds (minimal, median and maximal forecasted spending). The cloud infrastructure deployment and an orchestration of the sorting software parts has been implemented with a popular devops tool called ansible. An automatic scaling has been achieved through a combination of open source monitoring software and callbacks on additional ansible scripts.  These scripts switched the additional virtual servers based on simple rules on CPU consumption boundaries on already deployed cloud infrastructure.


The successful cloud infrastructure integration helped to achieve accomplish a new  architecture that led to high elasticity and resiliency.  This helped the customer to react  to rapidly changing infrastructure project requirements during peak periods of the production e-commerce system (Black Friday). That led to new business opportunities  working collaboratively with different customer’s departments in electronic retail.