This section contains the call quality results experienced in this Intel case study. User reported results are discussed as well as management, reliability, and quality.

User Reported Results

Intel did not perform a formal MOS measurement with users. Feedback was collected through channels such as a weekly health check e-mail (yes/no voting buttons to indicate problems), a detailed qualitative user survey, and interviews conducted by a Human Factors Engineer.

Softphone with QoS users experienced good voice quality all the time, no matter what time of day whether in a conference call, outbound, or inbound call. These users were extremely satisfied with the quality of voice and in fact, with properly tuned audio settings on the laptops and good quality headsets, the voice quality was described as an improvement over their regular phones.

Voice traffic for hardware IP phone users had virtually no contention with the data traffic. These users also reported good quality voice. The make and model of hardware IP phones was important as we deployed a variety from multiple vendors. Users were dissatisfied with certain types of phones, so by trial and error Intel could determine which phones worked best in the environment. Once phones with reported issues were replaced, users described the quality of voice as very good at all times.

Where QoS capability between the access switches and the distribution switches was not available, we had some softphone users. These users expressed satisfaction with the quality of voice most of the time; however, they frequently experienced degraded quality of voice, reportedly several times per day. Often this was explained by the fact that heavy downloads were occurring on the access switches at the time of the calls. Without prioritization for voice traffic, call quality was impacted, even with the high availability of bandwidth in the LAN. This confirmed the necessity of enabling end-to-end QoS.

Users with legacy phones connected to the IP PBX through the PIMG also reported good voice quality. The only area that could be impacted through the voice quality plan was at the data center switch, where voice was prioritized and a separate VLAN was configured. The existing wiring was used between the PIMG and the legacy phones, ensuring dedicated connections.

Management, Reliability, and Quality

The reliability of the LAN was 99.98%. This was achieved through pro-active monitoring. There was no physical redundancy for the access switches. The traffic levels on the network were well understood through the baselining efforts and there were no bottlenecks identified. The utilization on the uplinks from the distribution to the core did not peak above 40% and the average utilization remained below 10%. Tools were in place to monitor the health of the network including visibility into the voice traffic that received priority. All management, monitoring, and alerting is centralized, providing end-to-end visibility into the network. From a management standpoint, the ability to push QoS parameters throughout the network from a central location is important in order to effectively manage a large deployment.