29-06-2012, 03:45 PM
Study of router off-loading effectiveness on power consumption and cost in a network
Study of router off-loading.pdf (Size: 401.37 KB / Downloads: 149)
Abstract
Effectiveness of the MPLS-TP based router off-loading
method is studied from power consumption and node
cost perspectives. We compare power consumption and
node cost between a conventional network and a
router-off loading network utilizing a reference network.
Introduction
The network traffic has been increasing due to the
emergence of new applications and services, resulting in
an increase in power consumption of network equipment,
while the revenue from the network has been decreasing.
Therefore, network operators must drive down both
OPEX and CAPEX in order to sustain network growth.
In the current networks, traffic traverses multiple router
hops, resulting in an increase in the number of router
interfaces and power consumption by the routers. One of
the effective solutions to these issues is to introduce a
router off-loading technique using sub-lambda switching
such as the MPLS Transport Profile (MPLS-TP)
technique, e.g. moving as much traffic as possible to the
lower layers [1-2]. In this paper, we study the
effectiveness of the router off-loading with the MPLS-TP
technique and Packet Optical Transport System (POTS)
which is emerging as a platform for router off-loading
technique in power saving and node cost reduction.
Explanation of router off-loading model
Figure 1 shows the definition of the node models. Figure
1 (a) represents a conventional model in which routers
switch all transit traffic. In this model, the WDM
equipment includes the OTN function, transponder, and
ROADM. Figure 1 (b) shows an MPLS-TP based router
off-loading model. The POTS includes the MPLS-TP
function, OTN function, transponder, and ROADM. This
model can bypass a router by switching transit traffic
utilizing MPLS-TP switching functionality.
Results of power consumption and cost analysis
For power consumption analysis, we used the NSFNET as
the reference network and the traffic demand matrix as in
[6]. We assumed each path was 10 Gb/s base and
determined by the shortest hop manner. To accommodate
the maximum traffic (e.g. 4 Tb/s in this case), every node
was assumed to have four shelves for the router and six
shelves for the WDM and the POTS equipment. The
obtained ratio of off-loading interface (L) was 48% in this
analysis. By using equations (1) and (2), the calculated
network power consumptions in the conventional model
and the router off-loading model were 448 kW and 403 kW,
respectively.
Conclusions
We studied effectiveness of the MPLS-TP based router
off-loading on power consumption and cost in a network.
In a case study, we used the NSFNET as the reference
network and showed the MPLS-TP based router
off-loading could save about 10% power consumption
and reduce about 45% node cost, compared to the
conventional method.