TY - GEN
T1 - Implementing data center network energy management capabilities in NS3
AU - Sondur, Sanjeev
AU - Ray, Madhurima
AU - Biswas, Joyanta
AU - Kant, Krishna
N1 - Funding Information:
ACKNOWLEDGMENT This research was supported by NSF grant CNS-1422921
Funding Information:
This research was supported by NSF grant CNS-1422921.
Publisher Copyright:
© 2017 IEEE.
PY - 2017/7/2
Y1 - 2017/7/2
N2 - In this paper, we present an enhanced and holistic energy model for the widely used Network Simulator, NS3. As computing becomes more energy efficient, data movement, and hence data center networks consume an increasing percentage of the total energy consumption and it is important to provide energy management capabilities in the network simulators. Our enhanced NS3 simulator supports the use of different sleep modes for each network port on a switch/router and endpoint, and backplane on switch/router in order to reduce the network energy consumption. It also supports two port level power management mechanisms - unidirectional (transmitter only) and bidirectional (transmitter and receiver). Moreover, the simulator supports local consolidation that consolidates traffic across multiple outgoing/incoming links into a node, and a global controller that monitors all links in the network via a SDN-like mechanism and helps make better local consolidation decisions. The implemented capabilities are illustrated by applying them to the popular fat-tree based data center network. It is shown that the local consolidation mechanism coupled with bidirection port-level energy management can reduce the network energy consumption substantially.
AB - In this paper, we present an enhanced and holistic energy model for the widely used Network Simulator, NS3. As computing becomes more energy efficient, data movement, and hence data center networks consume an increasing percentage of the total energy consumption and it is important to provide energy management capabilities in the network simulators. Our enhanced NS3 simulator supports the use of different sleep modes for each network port on a switch/router and endpoint, and backplane on switch/router in order to reduce the network energy consumption. It also supports two port level power management mechanisms - unidirectional (transmitter only) and bidirectional (transmitter and receiver). Moreover, the simulator supports local consolidation that consolidates traffic across multiple outgoing/incoming links into a node, and a global controller that monitors all links in the network via a SDN-like mechanism and helps make better local consolidation decisions. The implemented capabilities are illustrated by applying them to the popular fat-tree based data center network. It is shown that the local consolidation mechanism coupled with bidirection port-level energy management can reduce the network energy consumption substantially.
UR - http://www.scopus.com/inward/record.url?scp=85051023686&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85051023686&partnerID=8YFLogxK
U2 - 10.1109/IGCC.2017.8323566
DO - 10.1109/IGCC.2017.8323566
M3 - Conference contribution
AN - SCOPUS:85051023686
T3 - 2017 8th International Green and Sustainable Computing Conference, IGSC 2017
SP - 1
EP - 8
BT - 2017 8th International Green and Sustainable Computing Conference, IGSC 2017
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 8th International Green and Sustainable Computing Conference, IGSC 2017
Y2 - 23 October 2017 through 25 October 2017
ER -
