Best Paper Award
Separating Predictable and Unpredictable Flows via Dynamic Flow Mining for Effective Traffic Engineering
Yousuke TAKAHASHI, Keisuke ISHIBASHI, Masayuki TSUJINO,
Noriaki KAMIYAMA, Kohei SHIOMOTO, Tatsuya OTOSHI,Yuichi OHSITA,Masayuki MURATA
[IEICE TRANS. COMMUN., VOL.E101-B, NO.2 FEBRUARY 2018]
Yousuke TAKAHASHI
Keisuke ISHIBASHI
Masayuki TSUJINO
Noriaki KAMIYAMA
Kohei SHIOMOTO
Tatsuya OTOSHI
Yuichi OHSITA
Masayuki MURATA
For Internet service providers to efficiently use network resources, they need to conduct traffic engineering (TE) to dynamically control traffic routes to accommodate traffic with limited network resources.
The performance of traffic engineering depends on the accuracy of traffic prediction.
However, the volume of network traffic has been changing drastically in recent years due to the growth of various types of network services, making traffic prediction increasingly difficult.
Our approach to tackle this issue is to separate traffic into predictable and unpredictable parts and to apply different control policies.
First, by flow-level traffic analysis, traffic is separated into two types of flow groups: predictable macroflows whose traffic variation is small and unpredictable macroflows whose traffic variation is large.
Furthermore, each separated traffic is controlled with a control policy suitable for these characteristics: predictable macroflows are controlled by prediction-based routing which calculates the route for resource optimization based on the predicted traffic volume, and unpredictable macroflows are controlled by load-balanced routing for congestion avoidance according to the remaining bandwidth of the network.
An evaluation of actual traffic measured in an Internet2 network shows that compared with current TE schemes the proposed scheme can improve the utilization efficiency of network resources and reduce the risk of network congestion.
As described above, this paper proposed a novel approach to separate traffic according to its characteristics and to control each type of traffic with a control policy suitable for these characteristics in order to overcome the limitations of current TE schemes. Moreover, through simulations conducted with actual traffic traces, we demonstrated that the proposed scheme can both optimize the performance of predictable traffic and enable robust control for unpredictable traffic.
The results of this paper suggest that the new TE scheme based on traffic characteristics analysis instead of current TE scheme based on traffic volume prediction is useful.
