We wanted to ensure that our product was accessible under heavy workloads, which meant going for the cloud scalability. We also wanted to implement Big Data/ AI/ ML algorithms instead of basic scripts we had at that moment, to ensure the system works better the more it is used.
We wanted our site to be deployed at AWS and use AWS CloudFront as CDN and AWS ML/AI services and Amazon RDS for back-end operations. However, IT Svit has calculated some expense projections and showcased that a bespoke Big Data solution would be more cost-efficient.
The main benefits they proposed were the significant decrease in false-positive error reports due to the correct ML model training, as well as the ability to launch testing in several browser-test instances simultaneously due to running them as Dockerized microservices atop a Kubernetes cluster.
IT Svit team has built a simple and elegant AWS infrastructure, using Amazon EC2 instances, S3 storage, Kubernetes cluster and Docker test-instances, PostgreSQL as the main database and Amazon RDS for scalability. Instead of using AWS AI/ML services, they deployed and trained a bespoke AI algorithm in Python, which checked the website snapshots before and after the release and highlighted the differences and errors. Review collected by and hosted on G2.com.