Simple Network Performance Detection

Testing download speed can only give us a rough picture of network efficiency. Sometimes we need to collect more information.

1: Test Response latency and Packet Loss Rate

Ping-l package size Application Server IP-N times (or-T)

For example, Ping 202.106.160.182-l 1024-N 300

This indicates that the local client can ping the application server 202.106.160.182, and ping the packet with the size of 1024byte. A total of 300 packets are pinged (if one packet is pinged in one second, in fact, the ping lasts for 3 minutes ). If-L is not specified here, the default value is 32 bytes. Sometimes we need to ping packets of different sizes multiple times, because the routes all have packet priority settings, sometimes the ping packet is not fast enough to see your network;

In addition to specifying packets of different sizes to Ping, you also need to consider pinging packets at different time points, because the Internet load has a certain regularity, such as am, am, PM, the online load is relatively large, and the ping results of these time points and other time points may be quite different.

Here is a ping result:

..........

Reply from 211.97.55.129: bytes = 32 time = 1921 Ms TTL = 244

Ping statistics for 211.97.55.129:

Packets: Sent = 300, received = 271, lost = 29 (9% loss ),

Approximate round trip times in Milli-seconds:

Minimum = 687 ms, maximum = 4233 ms, average = 1914 Ms

We can see that the response latency is very long and the packet loss rate is very high.

Response latency and packet loss rate. Any item can reflect the connection status.

2: Ping test maximum transmission unit

Sometimes, the packet loss rate of Ping packets is high, and the speed of the web pages on other websites is normal. In this case (especially in public network applications where VPN gateway is applied, check the MTU (maximum transmission unit) from the client to the application server or the tcp mss settings. Run the following command to check the MTU:

Ping-F-l packet size Application Server IP Address

-F indicates that the package size is not cut for transmission.

For example:

Ping-F-l 1472 10.1.70.1

Reply from10.1.70.1: bytes = 1472 time = 2 ms TTL = 255

Reply from10.1.70.1: bytes = 1472 time = 2 ms TTL = 255

......

Ping-F-l 1473 10.1.70.1

Packet needs tobe fragmented but DF set.

Packet needs tobe fragmented but DF set.

......

This indicates that the maximum transmission unit MTU from the client machine to 10.1.70.1 is 1472 + 28 (the length of the packet header encapsulated in the Ethernet) = 1500, and all packets larger than 1500 need to be cut. The default MTU in Ethernet is 1500. Therefore, the MTU in the client machine system and the MTU in the server system are generally 1500 by default, the MTU of related routes and gateways on the network is also 1500 by default, but in some cases, the default value may be less than 1500, especially for many VPN routing devices, if the MTU is inconsistent, the packet may be discarded, resulting in efficiency problems.

At this time, you need to consider adjusting the MTU of the client, server, route (or gateway) for testing until a suitable MTU value is found.

3: MTU and TCP MSS

There are many easy-to-use and intuitive tools on the Internet to optimize and adjust the MTU of the machine system, such as TCP optimizer:

You can easily modify the MTU value of the local machine.

It is convenient to detect the MTU from the client machine to the server.

For vro parameter settings, you can modify the MTU and tcp mss values of the vro interface to solve the problem.
The MTU and tcp mss values of specific packets should be tried based on the specific application and the best value should be selected based on the experience value. The MTU value can be selected for testing through the ping command; the selection of tcp mss values can be calculated by using MTU minus other encryption, link layer overhead, IP header, and TCP first-class bytes.

4: Route check:

Ping can collect the response delay, packet loss rate, and other information in the network connection. However, sometimes I need further details, such as the slow network, where is the slow speed, in this case, we can use commands such as tracert and pathping to obtain the slow route.

For example: tracert220.201.36.30
Tracing Route to 220.201.36.30 over amaximum of 30 hops

1 <10 MS <10 MS <10 MS 192.168.33.1 LAN address

2 <10 MS <10 MS <10 MS 192.168.101.1 LAN address

3 15 MS <10 MS <10 MS 10.14.27.1 LAN address

4 <10 MS <10 MS 16 MS 10.1.4.11 LAN address

5 <10 MS <10 MS <10 MS 10.1.10.33 LAN address

6 <10 MS <10 MS <10 MS 10.1.10.81 LAN address

7 <10 MS <10 MS 15 MS 218.61.254.173 Liaoning Netcom address

8 <10 MS 15 MS <10 MS 218.61.254.141 Liaoning Netcom address

9 16 MS <10 MS <10 MS 218.61.20.177 Liaoning Netcom address

10 <10 MS 15 MS <10 MS 218.61.20.197 Liaoning Netcom address

11 16 MS 31 MS 31 MS 219.158.8.241 Beijing Netcom address

12 15 MS 32 MS 31 MS 219.158.11.126 Beijing Netcom address

13 468 MS 454 MS 453 MS 219.158.28.214 Beijing Netcom address

14 453 MS 453 MS 453 MS 211.94.56.50 Beijing Unicom address

15 453 MS 453 MS 453 MS 211.94.54.214 Beijing Unicom address

16 484 MS 454 MS 484 MS 211.93.132.6 Liaoning Unicom address

17 469 MS 453 MS 469 MS 211.93.132.82 Liaoning Unicom address

18 469 MS 453 MS 453 MS 10.254.2.1 internal address of Liaoning Unicom

19 220.201.36.30 Destination Address

Trace complete.

From the above data, we can see that the network speed is mainly slow on the network interface 219.158.28.214 of Beijing Netcom and Beijing Unicom.