Cyber-attacks are deliberate attempts to gain unauthorised access to a person’s computer and data. As the web has grown, it has become a fundamental part of most people’s daily routine, and there are concerns about online security. In addition, the growing amount of information being shared between clouds and clients creates an attack surface that can be exploited. In addition to the increase in organizations and users, the attack surface has also increased. Deterioration in existing discovery plans is resulting in a reluctance to approve goals and an earlier acknowledgment of assaults. If no efficient assurance system is put in place, the web will grow increasingly susceptible, increasing the risk of information leakage. This research suggests an Intrusion Detection System (IDS) that divides packet traffic into attack and non-attack classes and uses Support Vector Machines to categorise malicious classes to identify network intrusions. The Support Vector Machine (SVM) is considered one of the most effective and promising techniques in machine learning Algorithms. On decision boundaries, Support Vector Machine Classification (SVMC) is founded. It supports all binary and multi-class classifications. KDDCUP 1999 and UNSW NB15 are two different cyber security datasets that we utilised in our research. Performance criteria have been used to assess the suggested model. The test results show that our methodology is more effective at identifying different cyberattacks. For the UNSW NB15 dataset and KDDCUP 1999 dataset, this model obtains remarkable attack detection rates of 99.8% and 98.2%, respectively.