Forecasting short-term data center network traffic load with convolutional neural networks

Abstract

Efficient resource management in data centers is of central importance to content service providers as 90 percent of the network traffic is expected to go through them in the coming years. In this context we propose the use of convolutional neural networks (CNNs) to forecast short-term changes in the amount of traffic crossing a data center network. This value is an indicator of virtual machine activity and can be utilized to shape the data center infrastructure accordingly. The behaviour of network traffic at the seconds scale is highly chaotic and therefore traditional time-series-analysis approaches such as ARIMA fail to obtain accurate forecasts. We show that our convolutional neural network approach can exploit the non-linear regularities of network traffic, providing significant improvements with respect to the mean absolute and standard deviation of the data, and outperforming ARIMA by an increasingly significant margin as the forecasting granularity is above the 16-second resolution. In order to increase the accuracy of the forecasting model, we exploit the architecture of the CNNs using multiresolution input distributed among separate channels of the first convolutional layer. We validate our approach with an extensive set of experiments using a data set collected at the core network of an Internet Service Provider over a period of 5 months, totalling 70 days of traffic at the one-second resolution.

Autonomous Virtual Machine Sizing and Resource Usage Prediction for Efficient Resource Utilization in Multi-Tenant Public Cloud

“…The authors report that AR-NN performs better than ARIMA on a number of accuracy metrics such as Mean Error (ME), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE). A similar conclusion has also been made in [34,35]. Neural network techniques have been successful in predicting cloud resource usage as seen in [36,37,38,39,40,41].…”

Prediction of Wave Power Generation Using a Convolutional Neural Network with Multiple Inputs

“…Linear functions used by the convolutional filters convert the input data into images in a sliding-window fashion [36]. Among the many deep neural networks, the CNN demonstrates excellent performance in the field of image processing, which comprises convolutional layers, pooling layer, and fully connected layers [37].…”

GCN-GAN: A Non-linear Temporal Link Prediction Model for Weighted Dynamic Networks

“…Several recent techniques have been proposed to predict the temporal links in dynamic graphs from different perspectives [2]- [5]. Despite their effectiveness, we argue that temporal link prediction remains a challenging task for two primary reasons.…”

“…For instance, the dynamics of communication links in ad hoc networks makes the design of routing protocol a challenging problem, where the prediction of the dynamic topology plays an important role to achieve a more efficient and reliable communication [1]. In data center networks, traffic prediction technique could be utilized to effectively schedule the highly parallel network flows while avoiding the performance degradation due to resource shortages [2]. For cellular networks, the prediction of users' locations can help to reduce the resource consumption (e.g., bandwidth) and achieve better Quality of Services (QoS) [3].…”

“…Although numerous studies have been developed to deal with the dynamics of network systems [2]- [5], most of them only focus on a specific application scenario (e.g., flow prediction in data center networks), failing to be generalized to other different scenarios.…”