Accurately predicting metrics such as bandwidth utilization in future networks can assist service providers in predicting network congestion, allowing for proactive network expansion, adjustments, and optimization. To adapt to the ever-changing network environment and requirements, methods for network traffic prediction have evolved from traditional statistical models to gradually incorporate Machine Learning (ML), Deep Learning (DL), and similar approaches. Given that real-world network traffic patterns are often nonlinear and have a long memory, DL algorithms like Recurrent Neural Networks (RNN) and Long Short-Term Memory networks (LSTM) are better suited for handling time series data. These algorithms excel in capturing the nonlinearity, long-term dependencies, and correlations among data points. In this paper, we outline an overview framework for Traffic Prediction (TP), encompassing problem definition, data collection, preprocessing, model selection, and model evaluation. We delve into the latest DL techniques in the field of network traffic prediction, highlighting the utilization of RNN, LSTM, and related models. Furthermore, we engage in a discussion of open research questions and provide insights into potential future directions for development.