In today's digital world, there is a very high amount of data which is stored on the clouds like Amazon Web Services, Google Cloud Platform, Jio Cloud or Azure, etc. As they have storing capabilities, there is also a risk for the data and mainly the attack called Distributed Denial of Service (DDoS) attack is commonly used to steal the data. This attack overloads servers by sending too many requests, causing them down for the other users. This research paper presents a real-time, scalable and low-cost system that can identify DDoS attacks on cloud instances. It builds using free-tier services of AWS like EC2 instances, CloudWatch and SNS services. It uses two EC2 instances - one instance acts as a server and one acts as a dashboard. Dashboard tracks real-time traffic, packets and CPU usage and displays suspicious requests. When CPU utilization crosses a certain limit (50%) the system will send the email and stop the server to prevent the system from DDoS attack. This system can be improved with machine learning algorithms to detect malicious IPs automatically. It currently uses models like Decision Tree, Isolation Forest, K-nearest-neighbor, Naive Bayes to detect DDoS traffic. Decision tree is used here to detect DDoS attacks; it gives 98.10% accuracy. Isolation forest performs well with 95.55% accuracy. KNN gave 80% accuracy and Naive Bayes gave 72.75% accuracy. The system also has useful features like geo-location. It helps to detect the location of the traffic, from which country and city, traffic is coming. The dashboard gives a clear view of Network traffic and packet analysis, request type, CPU utilization. It also evaluates whether the IP is safe or harmful. This paper helps to grow in the future for building an advanced system with automatic detection using artificial intelligence.