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.

Fig 1. Typical convolutional network architecture for image recognition.

Fig 2. Our convolutional network architecture for time series forecasting.

Fig 3. The time series from a weekend in February.

Fig 4. The time series from five weekdays in March.

Fig 5. ACF (left) and PACF (right) of the raw data.

Fig 6. ACF (left) and PACF (right) after first difference.

Fig 7. ACF (left) and PACF (right) of the raw data at the 64-second resolution.

Fig 8. ACF (left) and PACF (right) after first difference at the 64-second resolution.

Fig 9. Model tests (February, March and April).

Naive, ARIMA and ANN and CNN with multiresolution context. Weekends.

Fig 10. Model tests (June and July).

Naive, ARIMA and ANN and CNN with multiresolution context. Weekends.

Fig 11. Model tests (February, March and April).

Naive, ARIMA and ANN and CNN with multiresolution context. Weekdays.

Fig 12. Model tests (June and July).

Naive, ARIMA and ANN and CNN with multiresolution context. Weekdays.

Fig 13. Model durability tests.

Naive, ARIMA and ANN and CNN with multiresolution context. Weekends.

Fig 14. Model durability tests.

Naive, ARIMA and ANN and CNN with multiresolution context. Weekdays.