**6.4.1 SLA verification in centralized architecture:**

Case 1: The QoS offered to Alice and Bob matches SLA contract, perceived video quality is satisfying (Figure 15).

Fig. 15. Video bandwidth =128kbps.

Case 2: the network conditions, hence video quality, deteriorate proportionally to mass of competing services for lost packets and reduced flow rate (Figure 16).

Fig. 16. Video bandwidth =76kbps.

Case 3: competing services overload the routers: queues fill in gateways, impacting delay and jitter. Routers discard packets in excess, this causes static pixels in video (Figure 17) and in some cases service cancellation.

eTOM-Conformant IMS Assurance Management 71

plays a vital role in reduction of complexity and processing, which highlights the advantage of using TCP /IP for exchange parameters of service and performance indicators and

The response time of WS-synchronization and other agents resource is short compared to WS-Resource in SLA verification, because of the processing performed locally in each device, that has a potential to reduce traffic between Monitoring and Assurance layers, as well as it reduces

transmission at higher levels in order to achieve continuous monitoring.

response time compared to centralized architecture in tree cases (Figure 19).

Fig. 19. Response time (ms) for centralized and disturbed architecture.

in terms of CPU and memory.

QoS or failure of an entity.

Assurance and Fulfilment.

a QoS degradation and SLA violation (Figure 20).

**6.4.3 SLA violation** 

The processing of parameter flow in entity level allow a real-time control of multiple QoS and services, however a router must have enough memory for traffic conditioning, although using control function as treatment and comparison with thresholds can reduce its capacity

The distributed architecture provides a Monitoring layer that alerts Assurance layer within a very short time or near real time, and allows rapid processing of SLA violations, compared to another architecture where several treatments must be executed to detect degradation

Alice has registered in the IMS system with QoS classes Gold. The goal is to perform SLA Assurance tests in three representative cases and to compare results for SLA correction with

The MOS-AV (Mean Opinion Score-AudioVisual (BELLAFKIH et al., 2010) is a quality indicator for detecting user satisfaction requiring a video flow such as VoD (Video on demand) or IPTV (IP Television). It is a quantitative indicator taking values between 0 and 5. The MOS-AV reflects user satisfaction and SLA violation in our case the value 2.5 represent

Fig. 17. Video bandwidth =40kbps.

The platform succeeds in identifying accurately deterioration of delivered services. The cost in terms of response time has been evaluated as well. It is observed that response time for Resource-WS is much longer than for other web services, due to complexity of its tasks (Figure9, 10).

Fig. 18. Response time for different web services (Client: Alice).

The number of web services and their internal functionality has a considerable impact on running time of SLA verification. This led to limit the exposed eTOM processes to level 3 and to implement sub processes via internal java methods.

#### **6.4.2 Centralized vs distributed architecture**

The execution time in SLA verification is composite, and is directly related to processing time in each WS. This time varies depending on the number of planned operations, WS state and SLM&M conception. Similarly, nature of communication technology between entities

The platform succeeds in identifying accurately deterioration of delivered services. The cost in terms of response time has been evaluated as well. It is observed that response time for Resource-WS is much longer than for other web services, due to complexity of its tasks

The number of web services and their internal functionality has a considerable impact on running time of SLA verification. This led to limit the exposed eTOM processes to level 3

The execution time in SLA verification is composite, and is directly related to processing time in each WS. This time varies depending on the number of planned operations, WS state and SLM&M conception. Similarly, nature of communication technology between entities

Fig. 18. Response time for different web services (Client: Alice).

and to implement sub processes via internal java methods.

**6.4.2 Centralized vs distributed architecture** 

Fig. 17. Video bandwidth =40kbps.

(Figure9, 10).

plays a vital role in reduction of complexity and processing, which highlights the advantage of using TCP /IP for exchange parameters of service and performance indicators and transmission at higher levels in order to achieve continuous monitoring.

The response time of WS-synchronization and other agents resource is short compared to WS-Resource in SLA verification, because of the processing performed locally in each device, that has a potential to reduce traffic between Monitoring and Assurance layers, as well as it reduces response time compared to centralized architecture in tree cases (Figure 19).

Fig. 19. Response time (ms) for centralized and disturbed architecture.

The processing of parameter flow in entity level allow a real-time control of multiple QoS and services, however a router must have enough memory for traffic conditioning, although using control function as treatment and comparison with thresholds can reduce its capacity in terms of CPU and memory.

The distributed architecture provides a Monitoring layer that alerts Assurance layer within a very short time or near real time, and allows rapid processing of SLA violations, compared to another architecture where several treatments must be executed to detect degradation QoS or failure of an entity.
