One of the most important tasks of effective network management is identifying and classifying traffic flows at the network level. Because encryption makes it difficult for network devices to identify the applications running. Existing techniques are likely to become ineffective in terms of performance because of the proliferated usage of encrypted communication. We propose a model called ‘EnsLearn’, that efficiently identifies encrypted traffic to address these challenges. This approach presents a technique that increase accuracy when applied to online encrypted traffic. Applications that run on the internet use reliable encryption methods to secure communication. These encryption techniques guarantee that no other party may reveal the data. As a result, this study determines whether traffic is encrypted or not. The zenodo dataset is used to train the model. The Wireshark application is used for collecting online data for testing. Initially, the dataset is divided into 64, 128, 256, and 512 KB files. Each one is segregated into binary, MP3, text, video, PDF, and image files. These files are converted into comma-separated values (CSV) for feature extraction using CICFlowmeter and apply preprocessing steps to make the dataset into compatible form. Next, dimensionality reduction techniques FastICA and IPCA are used for feature selection. Gradient Boosting, CatBoost, and AdaBoost are used for classification. Experiments show that the best results are achieved with the proposed approach CatBoost with dimensionality reduction technique IPCA as its classification model, where it achieves an accuracy of 99% in 0.10 seconds for encrypted traffic classification.