Page 195 - Big data - Concept and application for telecommunications

Big data - Concept and application for telecommunications 5

Big-data-driven networking (bDDN) solves this problem by make use of big data generated by the network
itself. The bDDN separates complex data computing and processing functionalities from the network control
plane and management plane, and converges it into a big data plane. This big data plane has powerful big
data computing and analytical ability, can perform pervasive and inclusive network data collection and
computation, and extract useful information and intelligence from the big data. By applying big data
technologies to massive data in the future network, the bDDN provides computational data intelligence
support for network management, operation, control, optimization and security, etc. Furthermore, the bDDN
would decrease the complexity of the network plane and management plane, which in return, would
facilitate smart management, improve user QoE, elastic expansion and easy adaptation to emerging business
requirements in the future network.

Figure 7-1 – The framework of big-data-driven networking (bDDN)

As shown in Figure 7-1, the bDDN framework is made up of three planes – big data plane, management plane
and network plane. The bDDN model differs from existing network models in two major ways:
1) The bDDN framework is tridimensional, unlike the traditional vertical layered model which focuses
on a common process of network traffic.

As time goes by, challenges are brought forward mostly by network measurement and management
issues. However, they are not clearly and independently illustrated in the traditional vertical layered
model. As a result, solving network problems in the traditional framework becomes increasingly
obscure and cumbersome. The bDDN model clarifies the three major facets of future networks, as
well as their relationships.
2) The new model introduces the big data plane to support the management plane and network plane.
The big data plane is introduced because we found neither the SDN architecture nor the traditional
network framework could handle big data challenges properly.
The SDN centralizes the problems described earlier and transfers problems to the network
management and control process. The burden is centralized and must be predigested there. Facing
"network inflation" in future networks, the complexity of network management would increase and
problems during the control process would burst out. Full intelligence is a must if future network
management is to solve problems one by one and to, step by step, diminish complexity.

Network and infrastructure 187