The lab exercise itself provided a foundation for creating a simulated environment that was focused on evaluating the impact of background link load on FTP traffic. After creating the initial simulation environment, we were able to validate the configuration by matching output data for FTP performance with the reference data provided in the IL_ABA instructions document. The graphs below illustrate the student lab configuration findings compared with the lab reference graphs. The graphs, while not an exact match, provide enough similarity to validate the student simulation environment.
Figure 1 – Point to Point Utilization (reference) Figure 2 – Point to Point Utilization (student) BACKGROUND LINK LOAD Impact on Email Traffic After validating the simulation environment is correctly configured, the lab Report Instructions ask us how background load affects network performance as it relates to email traffic and voice traffic. In order to assess this impact, it was necessary to add email and voice services to the Profile Configuration and the server named FTP located in the Washington DC subnet.
The graphs below illustrate the impact of background load on point to point throughput–> and mint to point utilization for the back_load and no_back_load scenarios. The data in these charts is reflective of using the predefined Email(heavy) application profile metric. Figure 3 – Email Point to Point Throughput (bits/sec) Figure 4 – Email Pont to Point Utilization It’s clear that background load has a significant impact on link throughput and utilization.
The simulation without background loading remains steady and relatively flat with throughput at roughly Kbps and link utilization at under 5%. When background load is added, we see a marked upward trend in utilization ND throughput early in the simulation that keeps with our scheme of incrementally ramping up background load from 19,200 to 32,000 during the first 8 minutes, followed by a less pronounced continued upward trend over the remainder of the simulation.
Link throughput begins to stabilize at just over Kbps late in the simulation, while link utilization approaches 50%. Interestingly, we see a sharp drop in email download response time during the first seconds of both scenarios even as background load is ramping up at the same time. However, both scenarios flatten out to a more consistent level as the emulation progresses. The background load simulation stabilizes at roughly 1. 4 seconds for email download response time while the no background simulation settles at about . 7 seconds (see graph below).
Figure 5 Email Download Response Time (sec) Impact on Voice Traffic As with the FTP and email simulations, the topology was updated to include Voice Over IP (PC Quality) followed by running new simulations for both scenarios. In this run, we continue to look at point to point throughputј and point to point utilizationј. Additionally, we measured packet end to end delay (seconds). Without the use of other voice configuration metrics such as code selection and quality of service for voice packets, in either scenario voice over IP fails as a result of the high packet delay.
The chart below illustrates that the no background load simulation provides better performance for voice packets with an average delay of about 7 seconds. Alternatively, the simulation including background load produced an average packet delay of roughly 8 seconds. Figure 6 – Voice Packet End to End Delay (sec) When looking at link utilization metrics, on the other hand, there does not appear to be a significant difference between the background load simulation and the no background load simulation.
Figure 7 below indicates that, with the exception of the simulation startup being pre-loaded with 19,200 Kbps in background load, both scenarios show a sharp ramp up over the first 8 minutes before leveling off at just over Kbps. Link utilization, illustrated in figure 8, produced similar results in terms of the trend lines between the two scenarios. Both ramp up sharply over the first several minute and begin to level off as the link utilization approaches 100%.
With utilization so high, it’s clear that this particular implementation is not optimized for convergence. Voice services alone would consume all available bandwidth between East Coast sites making it impossible to support email and FTP services concurrently. Figure 7 – Point to Point Throughput (bits/sec) Figure 8 – Point to Point Utilization CONCLUSION Beginning with the initial lab exercise of comparing throughput and link utilization for FTP stabilized at roughly 10% of capacity while adding background load resulted in a peak utilization of about 55%.
Similarly, link throughput for mail remained under Kbps and 5% link utilization with no background load present while spiking up to nearly 50% of link capacity when background load is added. Finally, we learned that there is still some optimization work that needs to be done in order for this topology to be ready for convergence. Both scenarios resulted in nearly 100% link utilization when voice over IP was the only supported service. In any event, it’s clear that background load has a significant impact on the user experience based on download response times as well as efficient use of available link bandwidth.