Performance
- 4. Performance
- 5. Wrap Up
4. Performance
The numbers above show best case performance about equal to HomePNA (phoneline) 2.0 products. Considering that I've only tested HPNA products with a stand-alone 50 foot phone extension cable and not with actual home wiring, these results are pretty impressive! (Remember that the first generation powerline networking products were lucky to run at about 100 kbps... if they ran at all!)
Since this was my first crack at looking at HomePlug networking, I thought I'd run some longer tests with NetIQ's Chariot, and generate some comparative plots. The results are shown in the plot (click on it for a full-sized view) and table below.
(Test conditions are plotted in order, starting with Condition 1.)
| Test Condition | Average (Mbps) | Minimum (Mbps) | Maximum (Mbps) |
Throughput 95%
Confidence Interval |
Measured Time (secs) |
Relative Precision
(smaller number is better) |
| Condition 1 | 5.814 | 1.240 | 6.299 | 0.099 | 59.723 | 1.710 |
| Condition 2 | 5.225 | 0.954 | 5.797 | 0.173 | 59.710 | 3.312 |
| Condition 3 | 5.245 | 1.060 | 5.839 | 0.140 | 59.637 | 2.661 |
| Condition 4 | 2.933 | 2.564 | 3.089 | 0.010 | 59.743 | 0.329 |
The results show that the throughput is pretty consistent over the one minute test period, even with the large negative "spikes" that you see on the plots. I'm not sure what the "spikes" are caused by, but I'm pretty sure that they're not due to any electrical interference on the line... at least not any that I was generating! Condition 2 and 3's measurements show about a 10% throughput loss, and Condition 4 had a drop of just about 50% exactly. I can't explain the difference between the Condition 3 and 4 results, since both were on the other phase of my home's electric service and roughly equal in distance from the reference bridge in my office. I ran the tests a few times, and although the exact numbers varied, the relative throughput reductions in each of the test Conditions were consistent.
Ok, time for the torture test! I tried running both a hair dryer, then a small hand-held jigsaw while running a Condition 1 test to see if there was a throughput hit or a loss of connectivity. Both noise generators were plugged into the same outlet strip (not filtered) as both bridges, and I ran at least two tests with each appliance. The resulting plots didn't show any appreciable change in the throughput while the motors were running. Basically, the motor noise didn't make any difference in the network's operation... another impressive performance!
- Previous page Setup and Administration
- Next page Wrap Up
