Why do I get different results from different Internet Speed Test Sites?
The difference in results is due to the location of the Internet speed
test server you are testing with. Every test is subject to many factors
including, but not limited to, location of the test server (if you are
in Washington, D.C. and the test server is in Los Angeles, your results
will
be very different from a test you run to a server in the Washington,
D.C. area), testing method utilized (http, ftp, socket to socket,
etc.), speed of your Internet connection, software and optimization of
your computer, utilization of the remote speed test server and its
Internet connection, traffic congestion on the internet, time of day
attempted and other factors too numerous name here. If you are an
Atlantech Online Internet access customer, by using Atlantech's speed
test server, you eliminate traffic congestion on the internet as a
factor since your Internet service and the server we use are on the
same provider network.
Why are download and upload speeds different? The
server you are connected to perform the speed test is optimized for
sending data (speed tests up to 100Mbps have a high level of accuracy).
The server's capacity for uploading data allows for upload tests of up
to 10Mbps at a high level of accuracy.
Why do I get different results at different times? Under
ideal network conditions, Atlantech's Speed Test Server will provide
virtually identical results every time that it is run (verified in lab
conditions). And level of accuracy is possible because Atlantech's
Speed Test Server measures data transfers down the millisecond (1/1000
second), while most other speed testers can only time data transfers
accurate to 1/100 of a second (10's of milliseconds). So Atlantech's
Speed Test Server is very precise.
However, in most environments, users share the network, Internet
connection, and the destination server, which can cause a little
variation in the results.
If you see wildly different results each time Atlantech's Speed Test
Server is run, that strongly suggests that you are sharing your network
and/or server with too many other Internet users. To obtain a the most
accurate results, you must remove all devices from the internet
connection other than the device you are using for the speed test.
Why is a direct socket-to-socket connection test more accurate than an HTTP test?
The difference in results is due to the different layers of the OSI model.
A direct socket-to-socket connection works at the lower TCP transport
layer. A HTTP connection uses a browser which works at the application
layer and is subject to factors such as browser caches and proxies
which can affect results, particularly on high speed networks (see the
following FAQ).
What speed test results can be expected?
You can set the download and uploads tests independently to use HTTP or
socket testing. The socket option provides the most accurate speed
tests available (see previous FAQ). When socket tests are specified,
the standard HTTP speed test is used as a backup in case a client's
firewall prevents direct socket connections to the Atlantech's Speed
Test Server's ports.
RATED NETWORK SPEED
EXPECTED SOCKET RESULTS
Network
download
upload
download
upload
LAN
10 Mbps
10 Mbps
9.49 Mbps
9.49 Mbps
LAN
100 Mbps
100 Mbps
94.9 Mbps
94.9 Mbps
HTTP Speed Tests: Atlantech's Speed Test Server
measures actual HTTP data transfer performance -- not rated network
speeds. Since there will always be some overhead associated with a
network connection, you can expect results on high quality networks
(and reasonably fast computers) to approach around 95% of the rated
download speed and around 92% of the rated upload speed, similar to the
following:
RATED NETWORK SPEED
EXPECTED SOCKET RESULTS
Network
download
upload
download
upload
T1
1.544 Mbps
1.544 Mbps
1.46 Mbps
1.44 Mbps
Cable Modem
3.0 Mbps
256 kbps
2.88 Mbps
241 kbps
LAN
10 Mbps
10 Mbps
9.49 Mbps
9.15 Mbps
When Atlantech's Speed Test Server was tested in a local 100 Mbps LAN
environment, with a client connected to the server via a 10/100 Mbps
switch with various speed and duplex settings, the following
'download/upload' results were observed (all numbers are in Mbps):
10 Mbps Full
10 Mbps Half
100 Mbps Full
100 Mbps Half
9.49/9.17
8.08/7.77
85.4/69.0
70.4/65.1
9.49/9.17
8.12/7.98
82.1/70.4
71.4/65.1
9.49/9.10
8.12/7.65
82.6/69.0
71.7/66.3
9.49/9.17
8.05/7.70
85.9/69.0
71.2/66.3
9.49/9.12
8.11/7.77
84.9/71.8
72.6/62.7
Please note: (1) The CPU speed of the server and/or
client used can dramatically affect the results. (2) The upload results
for 100 Mbps are displayed for reference only - even though we only
rate Atlantech's Speed Test Server HTTP upload accuracy to 10 Mbps.
Upload: Multiple upload sizes, ranging in size from 8 KB to 128 KB.
What is jitter? Jitter is the variation in time between packets sent
and packets arriving caused by network difficulties such as route
changes, congestion, packet loss, traffic regulators etc., and plays a
major role in the quality of a VoIP call. VoIP works by sending voice
data as a stream of packets from source to destination. These packets
can take a varying amount of time to reach the destination and
invariably do not arrive in the order in which they were sent.
For a VoIP telephone call to work well the packets sent from the source
must arrive within a certain time window (or ‘buffer') in order for
the receiving end to reassemble the packets in the correct order and
reproduce the spoken words. When there is excessive jitter the time
delay is too long (high latency) and packets arrive outside the time
window and get lost from the call (discarded). As a result, the
recomposed sound no longer reflects exactly what was sent, and
depending of the extent of the delay may not be understandable by the
recipient.
What is packet loss? Packet loss plays a key role in the quality of VoIP
connections, as high packet loss causes some of the voice data not to
arrive to the recipient. Packet loss occurs when voice packets are
discarded by the jitter buffer (see previous FAQ), or dropped by
network routers/switches due to high congestion. Atlantech's Speed Test
Server measures the percentage of packet loss and reports the
associated level of sound quality.
What is MOS?
The MOS, or Mean Opinion Score,
is a numeric measure of the sound quality at the receiving end of a
communication circuit. Although the score is subjective it provides a
widely-used method to rate the quality of voice communication in a
simple way that meaningful to end users. The score is normally between
1 and 5 with 5 being the best.
The MOS value is reported in the Atlantech's Speed Test Server summary
tab once a connection test completes, a VoIP simulation that drops
below 3.5 is considered poor quality, a measure of 4.2-4.5 is
considered good quality.
What is the Quality of Service (QOS) percentage?
Atlantech's Speed Test Server reports the Quality of Service (QOS)
for your Internet connection, which represents the level of consistent
download bandwidth (or throughput) provided by Atlantech (assuming
Atlantech is your ISP). It is the ratio of the slowest download rate to
the highest download rate, as seen during a single Atlantech's Speed
Test Server download test.
The higher QOS percentage, the higher the overall quality of the
connection -- which results in better connections for VoIP, video and
bandwidth intensive applications.
What is Max Pause? Max Pause is the longest pause recorded during the
download speed test. This should be a very small number. If not, it
could indicate Internet congestion or a bad Internet connection.
What is RTT? Round-Trip Time (RTT) is the time taken for
Atlantech's Speed Test Server to send a small request packet to the
server, and to receive back a response. This indicates how good the
connection path is between you and the server. For best results on a
broadband connection, a connection should have a RTT of less than
150ms, and ideally less than 100ms.
