As　the　Internet　and　big　data　technologies　advance,　a　tremendous　amount　of　data　is　generated　daily.　Efficient　network　operations　are　essential　for　handling　such　data.　Accurately　predicting　future　network　traffic　in　real-time　enables　prompt　response　from　cloud　infrastructure　and　efficient　traffic　scheduling　and　allocation,　ultimately　reducing　costs,　preventing　economic　losses,　and　optimizing　the　performance　of　downstream　facilities.　However,　predicting　network　traffic　in　large-scale　data　centers　is　challenging　due　to　the　multidimensional,　nonlinear,　and　high-volatility　nature　of　the　time　series.　Traditional　prediction　methods,　e.g.,　regression　algorithms,　struggle　to　capture　non-linear　features　effectively.　Many　deep　learning　models　face　issues　such　as　gradient　explosion　or　vanishing　during　their　training.　Current　commonly　used　prediction　methods　fail　to　fully　uncover　vital　information　about　the　frequency　domain　features　in　the　time　series.　To　do　so,　this　work　proposes　a　novel　network　traffic　prediction　model　that　combines　a　Savitzky　Golay　filter,　sequence　decomposition,　multi-scale　attention,　a　temporal　convolutional　network,　an　autocorrelation　mechanism,　and　bidirectional　long　short-term　memory.　It　can　accurately　predict　future　network　traffic　by　adaptively　extracting　important　features　without　requiring　excessive　feature　engineering　of　the　original　data.　It　performs　end-to-end　sequence　prediction,　feature　selection,　and　automatic　learning　of　data　features　and　temporal　dependencies　to　achieve　accurate　time　series　prediction　and　avoid　redundant　data　processing.　Experiments　involving　ablation　studies　and　comparisons　with　advanced　prediction　models　are　performed　with　the　Google　cluster　trace.　Experimental　results　show　that　the　proposed　model　improves　the　prediction　accuracy　by　at　least　38.52%　over　the　state-of-the-art　models.　©　2024　IEEE.