This visualization shows loss of UDP packets from the client to the
server (upstream) or from the server to the client (downstream)
measured in percentages. Packet loss is applicable to UDP tests
only, because in TCP, all packets must be acknowledged and no data
loss may occur. UDP loss is calculated as the percentage of data
that was lost during transmission. For example, if the server sent
1 megabit of data in 10 milliseconds and the client received 0.6
megabits in 10 milliseconds, while 0.4 megabits were lost en route,
a 40% downstream loss has occurred.
UDP loss determines end user experience in audio and video
streaming applications, such as VoIP. High loss percentage might
cause high jitter and delays in audio and video.
When viewing this visualization, it is very important to understand
high downstream loss is normal.
UDP traffic is not acknowledged. This means that the party that
sends traffic can send as much traffic as the networking system can
handle without “caring” about how much of it will be lost. A
typical computer on the wired side of the network (server) equipped
with a gigabit adapter can send hundreds of megabits per second.
This data will first reach a switch, which might be the first
bottleneck, and then the AP, which is almost always a bottleneck,
because a typical 802.11n access point cannot send more than 100 or
150 Mbps of data downstream, i.e., to the client. As a
of UDP packets might be lost en route, but this is the only way to
figure out the maximum downstream UDP throughput value.
When high UDP loss areas are discovered, the following solutions
Verify that the actual PHY rate is sufficient. Throughput rates
cannot exceed PHY rates; in practice, they are about 50% lower than
PHY rates. For example, if the PHY rate in the given area is only 2
Mbps, you should not expect the throughput rate to exceed 1 Mbps.
In practice, it might be as low as 0.1 or 0.2 Mbps, depending on
Other common reasons for low throughput rates are interference and
excessive network traffic. The Signal-to-Interference Ratio visualization
available for passive surveys might provide insight into
interference issues. Excessive network traffic might be caused
either by oversubscription (too many clients per AP) or by
excessive network load by some clients. The former can be addressed
by increasing the number of APs, while the latter should be
verified and handled by network traffic monitoring software.