Page 445 - Handbook of Modern Telecommunications

3-236 CRC Handbook of Modern Telecommunications, Second Edition

3.9.4 Selecting and Applying Sensory Systems

Each of the sensory technologies we have covered has advantages and disadvantages. While agent-based
and device-based sensory technologies contribute valuable information to provider operations, due to
their limited viewpoints they can only play a secondary role in sensory network architecture. Similarly,
CDR-based sensory technologies operate in a retrospective manner, making them valuable for offline
capacity planning and service product planning, but less useful for real-time operations and prob-
lem troubleshooting. Probe-based systems for signaling or packet-flow monitoring provide extensive
data, but cannot tell you what is happening inside the service delivery network elements themselves.
Consequently, a mix of these technologies will be required to meet all potential operational needs, and
should be selected based on current gaps, operational priorities, and service objectives.
3.9.4.1 Visibility
The primary function of a sensory network is to provide a clear image of the health and activity in a
communications delivery infrastructure. Importantly, while detailed and somewhat unique views and
presentations of sensory data are needed to satisfy each of the many functional areas described above,
the sensory network technology and architecture decisions can be driven by focusing on three key view-
points—network-centric, service-centric, and customer-centric [STRA07].
3.9.4.1.1 Network Visibility
In a circuit-switched network, there are predefined channels (circuits) that carry the message depend-
ing on the destination, with each channel operating at a fixed bandwidth. Regardless of message size,
information is transported to its intended destination at the allocated transport level. For IP, bandwidth
allocation is variable according to the number of messages to be processed (incoming or outgoing).
In the PSTN, an operator monitors the percent utilization of each channel so it knows when a chan-
nel is filling up and more bandwidth is needed. Traffic on an IP network often fluctuates, and conten-
tion issues arise when there are more messages than bandwidth, causing new messages to be refused or
blocked. If a message is blocked, the network is designed to retransmit the message, however the retrans-
mitted message can also be blocked if bandwidth still is not available. This process can be repeated sev-
eral times resulting in message delays that often generate customer complaints.
IP network virtualization presents a different problem as providers migrate toward a more integrated,
if not entirely converged, service offering environment. Virtual IP service networks can include the net-
work itself (e.g., virtual network operators), the data storage layer, any number of network-supplied ser-
vices (e.g., VPN, virtual voice network), and the application delivery layer (e.g., Web services, SLAs) to
name a few. These multiple abstraction layers make it difficult to determine, for example, the root cause
of a reported problem or to measure the quality of service (QoS) associated with a customer offering.
Virtualization problems are not visible through device-based sensory monitoring, signaling sensory
monitoring, agent-based sensing, or CDR-based sensing—only packet-flow sensory systems have the
depth of detail to see through the abstraction fog of virtualization and help deliver understanding of the
flow of different services and service components across complex IP networks.
3.9.4.1.2 Service Visibility
Services can take on many forms including pure network-based services such as voice mail, e-mail,
three-way calling, and broadband access. They can also involve combinations such as broadband access
with third-party applications or digital content. Due to the number of potential network and partner-
supplied capabilities that can be used to define a service, and the eventual goal of delivering services
targeted to each individual customer-subscriber, assuring such combinations is a dynamic and some-
times monumental undertaking. To create an integrated view of service quality, and to monitor service
levels experienced by individual subscribers, a number of service quality indicators defined around each
component of a service should be considered. For example: