Using Global Accelerator to Speed Up Cross-Border Access to Web Servers

Overview

Application scenario: If you deploy your web server in the Chinese mainland, users outside the Chinese mainland may face problems such as high latency, packet loss, and jitter, due to unstable cross-border networks. To address these issues, you need a global accelerator.

Solution architecture: Suppose you have a web server deployed in CN East-Shanghai1 and your domain name has been licensed and mapped to the EIP in CN East-Shanghai1 bound to the server. Users can access your website using the domain name over the Internet. To accelerate cross-border access to your website, you can use DNS to map your domain name to the anycast IP address of a global accelerator, so that users across the globe can access your website faster through the Huawei backbone network.

Resource and Cost Planning

Flowchart

Step 1: Apply for a cross-border permit.

In accordance with the laws and administrative regulations of the Ministry of Industry and Information Technology (MIIT) of the People's Republic of China, only China Mobile, China Telecom, and China Unicom are allowed for cross-border network communications, and a cross-border permit is required if you carry out business activities outside the Chinese mainland.

1. Click Request a Cross-Border Permit.

   The Cross-Border Service Application System page is displayed.

   Figure 1 Applying for a cross-border permit
   
2. On the application page, set related parameters and upload related materials.
3. Click Submit.

Step 2: Buy a global accelerator.

To use Global Accelerator for faster access, you first need to create a global accelerator.

1. On the Global Accelerator page, click Buy Global Accelerator.
   Figure 2 Buying a global accelerator
   
2. Set parameters. Select Outside the Chinese mainland for Applicability. For other parameters, see Table 2.
   Figure 3 Creating a global accelerator
   

   Table 2 Parameters for configuring a global accelerator

   Parameter	Description
   Name	Name of the global accelerator you want to create. Only letters, digits, and hyphens are allowed. You can enter up to 64 characters.
   Enterprise Project	An enterprise project you would like to use to centrally manage your Global Accelerator resources. You can use an existing enterprise project or create one.
   Applicability	Where the global accelerator will be used. There are two options: Outside the Chinese mainland or Chinese mainland. Outside the Chinese mainland is selected by default. Outside the Chinese mainland is recommended for this practice.
   IP Address Type	The type of the IP address used by the global accelerator. IPv4 is selected by default.
   Tags	An identifier of the global accelerator. Each tag consists of a key and a value. You can add 20 tags for a global accelerator. NOTE: If a predefined tag has been created in TMS, you can select the corresponding tag key and value. For details about predefined tags, see Predefined Tag Overview. If you have configured tag policies for Global Accelerator, you need to add tags to your accelerators based on the tag policies. If you add a tag that does not comply with the tag policies, global accelerators may fail to be created. Contact the administrator to learn more about tag policies.
   Description	Supplementary information about the global accelerator. You can enter up to 255 characters.

3. Click Next.

Step 3: Add a listener to the global accelerator.

Add a listener to the global accelerator to route requests across endpoints based on the client affinity you set.

Configure the parameters as described in Table 3.

Figure 4 Adding a listener

Table 3 Parameters for configuring a listener

Parameter	Description
Name	Listener name. Only letters, digits, and hyphens are allowed. You can enter up to 64 characters.
Protocol	The protocol used by the listener to receive requests from clients. The protocol can be TCP or UDP.
Port	The ports or port ranges used by the listener to receive requests from clients. The port number ranges from 1 to 65535. You can enter one or more ports or port ranges separated by commas (,). Example: 1-10,11-50,51,52-200
Client Affinity	How requests are routed. There are two options: None: The listener routes requests evenly among the endpoints in the endpoint group. Source IP address (only for TCP and UDP listeners): The source IP address of each request is calculated using the consistent hashing algorithm to obtain a unique hash key, and all the endpoints are numbered and mapped to the hash keys. Requests from the same IP address are forwarded to the same endpoint for processing.
Tags	An identifier of the listener. Each tag consists of a key and a value. You can add up to 20 tags to a listener. NOTE: If a predefined tag has been created in TMS, you can select the corresponding tag key and value. For details about predefined tags, see Predefined Tag Overview. If you have configured tag policies for Global Accelerator, you need to add tags to your listeners based on the tag policies. If you add a tag that does not comply with the tag policies, listeners may fail to be created. Contact the administrator to learn more about tag policies.
Description	Supplementary information about the listener. You can enter up to 255 characters.

Step 4: Associate an endpoint group with the listener.

Associate an endpoint group with the listener in the CN East-Shanghai1 region and add an endpoint to this endpoint group as instructed by Table 4.

Step 5: Add record sets.

Add record sets to map your domain name to the anycast IP address of the global accelerator or the EIP bound to your web server.

This section uses Huawei Cloud DNS as an example.

1. Go to the Public Zones page.
2. On the Public Zones page, click the target domain name.

   The Record Sets page is displayed.

3. In the upper right corner of the page, click Add Record Set.
4. On Add Record Set page, add three record sets as instructed by Table 5.
   Figure 5 Adding an A record set
   

   Table 5 Parameters for configuring an A record set

   Parameter	Description
   Name	Prefix of the domain name to be resolved. For example, if the domain name is example.com, the prefix can be as follows: www: The domain name is www.example.com, which is usually used for a website. Left blank: The domain name is example.com. The Name field cannot be set to an at sign (@). Just leave it blank. *: The domain name is *.example.com, which is a wildcard domain name, indicating all subdomains of example.com.
   Type	Type of the record set. Select A – Map domains to IPv4 addresses for this practice.
   Line	Resolution line. The DNS server will return the IP address of the specified line, depending on where end users come from. Select Default, Region > Chinese Mainland, and Region > Global for the three record sets, respectively.
   TTL (s)	Cache duration of the record set on a local DNS server, in seconds. The value ranges from 1 to 2147483647, and the default value is 300. If your service address changes frequently, set TTL to a smaller value. Retain the default value for this practice.
   Value	IPv4 addresses mapped to the domain name. Set different values for the three record sets: If Line is set to Default or Chinese Mainland, set the value to the EIP of the web server. If Line is set to Global, set the value to the anycast IP address of the global accelerator.
   Weight	(Optional) Weight of a record set. The value ranges from 0 to 1000, and the default value is 1. Retain the default value for this practice.
   Tag	(Optional) Identifier of a record set. Each tag contains a key and a value. You can add a maximum of 10 tags to a record set.
   Description	(Optional) Supplementary information about the record set. You can enter a maximum of 255 characters.

5. Click OK.
6. Switch back to the Record Sets tab.

   View the record sets you have added and ensure that their status is Normal.

   

Verifying Acceleration

You can run the curl command on a Windows PC in the area where acceleration is required to check whether the access is accelerated.

1. Open the cmd window and run nslookup <Website domain name > to check whether the anycast IP address is returned.
2. Run the following command to check the latency of accessing the EIP in CN East-Shanghai1 over the public network:

   curl -o /dev/null -s -w "time_connect: %{time_connect}\ntime_starttransfer: %{time_starttransfer}\ntime_total: %{time_total}\n" "http[s]://<IP>[:<Port>]"

   

   • IP: EIP bound to your web server.
   • Port: HTTP port number used by the web server.
   • time_connect: time taken to establish a TCP connection, in seconds. It is from the time when a TCP connection request is initiated to the time when the connection is established.
   • time_starttransfer: time when transfer starts, in seconds. It is from the time when the client sends a request to the time when the endpoint replies with the first byte.
   • time_total: total connection time, in seconds. It is from the time when the client sends a request to the time when the endpoint responds to the request.
3. Run the following command to check the latency of accessing the anycast IP address:

   curl -o /dev/null -s -w "time_connect: %{time_connect}\ntime_starttransfer: %{time_starttransfer}\ntime_total: %{time_total}\n" "http[s]://<IP>[:<Port>]"

   

   Set IP in the command to the anycast IP address provided by Global Accelerator.

4. Compare the values of time_connect and view the latency before and after acceleration.