Complementing, there is the possibility of doing this test using the Jmeter.
I will not go into details of how to use the same, only how to configure it to simulate different internet speeds.
Editing the file <jmeter_home>/bin/user.properties
and adding the properties:
httpclient.socket.http.cps=0
httpclient.socket.https.cps=0
Changes these properties to a desirable value, such as the examples below:
Mobile data GPRS : 171 kbit/s = 21888
Mobile data EDGE : 384 kbit/s = 49152
Mobile data HSPA : 14,4 Mbp/s = 1843200
Mobile data HSPA+ : 21 Mbp/s = 2688000
Mobile data DC-HSPA+ : 42 Mbps = 5376000
Mobile data LTE : 150 Mbp/s = 19200000
WIFI 802.11a/g : 54 Mbit/s = 6912000
WIFI 802.11n : 600 Mbit/s = 76800000
Ethernet LAN ; 10 Mbit/s = 1280000
Fast Ethernet : 100 Mbit/s = 12800000
Gigabit Ethernet : 1 Gbit/s = 128000000
10 Gigabit Ethernet : 10 Gbit/s = 1280000000
100 Gigabit Ethernet : 100 Gbit/s = 12800000000
WAN modems V.92 modems : 56 kbit/s = 7168
ADSL : 8 Mbit/s = 1024000
ADSL2 : 12 Mbit/s = 1536000
ADSL2+ : 24 Mbit/s = 3072000
For other values inform according to the formula:
cps = (target bandwidth in kbps * 1024) / 8
More details on official documentation.
All the answers are great and professional, but as far as I know the throttling of browsers does not simulate the actual latency of a slow network. It also does not simulate high packet loss rates. It wants to test the site with a real slow internet, it uses a 3G connection from Tim...
– Oralista de Sistemas
@Renan I’ve tested a lot this way, and I’m happy to say I don’t need it anymore these days xD... Now, about trolling, it doesn’t really simulate packet loss (or if it simulates I haven’t found it yet), but latency is configurable, as well as download and upload.
– Randrade