Organizations are increasingly adopting cloud computing services as an easily accessible, centralized and affordable substitute to internal IT systems. Similarly, today's globally competitive businesses use internet and networked applications to offer services efficiently and cost effectively but swift and agile e-marketing is a dual edged sword, alongside new business opportunities, high valued security threats and intrusions may breach important information. Intrusion detection has become a paramount issue when the big organizations adopted cloud computing for their important data storage on the cloud. Intrusions Detection Systems (IDS) are security programs to identify activities in the network that compromises the confidentiality, availability or integrity of resource. The primary goal of IDS is to identify malicious activities in the network traffic with high detection rate, low false alarms, lesser resources usage and minimum computational cost. Soft computing is an innovative field to develop intelligent IDS while minimizing the deficiencies in traditional as well as neural network IDS. The objective of this research is to propose an efficient soft computing approach with high detection rate and low false alarms while maintaining lower cost and lesser time. Our promising results using hybrid features selection approach LDA, GA named as GLDA (Genetic Linear Discriminant Analysis) and SVM (Support Vector Machines) Kernels as classifiers with different combinations of NSL-KDD datasets show that a new proposed system is an improved and applicable depiction of an ideal intrusion detection system.