What is Global Traffic Manager 3.0? - Alibaba Cloud DNS - Alibaba Cloud Documentation Center

Global Traffic Manager (GTM) 3.0 is a fully upgraded version of the GTM service. It provides a graphical interface for configuration orchestration, includes a rich set of health check templates, and supports multiple load balancing policies, such as proximity-based, weighted, sequential, and round-robin. GTM 3.0 also lets you combine multiple types of request lines. This service is designed to provide enterprises with a flexible and efficient disaster recovery scheduling mechanism for fine-grained and intelligent management of inbound traffic.

Product overview

Global Traffic Manager (GTM) supports proximity-based access, high-concurrency load balancing, health checks, and failover. This helps enterprises quickly build active zone-redundancy and geo-disaster recovery architectures. GTM also supports the management of both Alibaba Cloud and non-Alibaba Cloud IP addresses. This allows enterprise customers to quickly build disaster recovery architectures for hybrid cloud applications.

GTM is a DNS-based service. It uses DNS to return specific endpoints to users. Clients then connect to these endpoints directly. Therefore, GTM is not a proxy, a gateway device, or an application access service. GTM does not process or view the network traffic between clients and application services. Click to purchase Global Traffic Manager.

After you activate a GTM instance, you must specify a CNAME endpoint. Then, you must add a CNAME record to map your business domain name to the GTM endpoint. This enables disaster recovery switchovers and intelligent domain name resolution for your application services.

Features

The following table describes the features of Global Traffic Manager.

Feature	Description	References
GTM endpoint	The endpoint of Global Traffic Manager (GTM) is the domain name through which GTM provides services. You typically add a CNAME record that maps your business domain name to the GTM endpoint to connect your business to GTM. The prerequisite for a GTM endpoint is that at least one independently hosted primary domain name or subdomain name exists under your account. For example, `example.com` is a primary domain name under your account. GTM endpoint format: `Host Record`.`Primary Domain Name`. Example: `gtm.example.com`. If your domain name is hosted on Alibaba Cloud DNS, you can enter the host record of your business domain name as the host record of the endpoint. In this case, your GTM endpoint is your business domain name.	Adding a domain name
Address pool	An address pool is a feature of GTM that is used to manage application service addresses, which can be IP addresses or domain names. An address pool represents a group of IP addresses or domain names that provide the same application service and have the same carrier or region properties. You can configure multiple address pools for a GTM instance. This way, users in different regions can access different address pools to implement proximity-based access. If an entire address pool becomes unavailable, a switchover to a backup address pool can be performed.	Address pool configuration
Address	An address is the endpoint of an application and also the response that is returned by GTM after the resolution and decision-making processes. An address can be an IP address or a domain name. The service port is an important parameter for performing health checks to detect service availability.	Address configuration
Load balancing policy	A load balancing policy is a dynamic resource scheduling mechanism that selects an appropriate address pool for an endpoint and an appropriate address within the address pool based on specific algorithms and policies. The following load balancing methods are supported: Proximity-based: This is an intelligent resolution feature that accurately identifies the source of a request based on the geographic location or network carrier of the client. This feature directs traffic to the endpoint that is geographically closest to the user or in the same carrier environment. This optimizes network latency and improves service response speed. Sequential: Address pools or addresses are selected one by one based on a preset sequence to ensure orderly access to available servers. Round-robin: All address pools and addresses are allocated in a round-robin manner to ensure that all available servers receive a relatively even number of access requests. This achieves an overall balanced distribution of loads. Specifically, a single address pool is selected from multiple address pools based on the round-robin mechanism. All addresses in the address pool are returned in a round-robin manner. Weighted (Weighted Distribution): Weight values are set based on the processing capabilities or performance of different servers. Addresses are allocated based on weight ratios. This allows servers with higher weights to handle more requests and achieves efficient utilization of system resources. These flexible load balancing policies allow for reasonable distribution and optimal management of inbound network traffic.	To meet complex scheduling requirements of enterprises, GTM 3.0 supports two-level access policy scheduling. The logic of the two-level access policy scheduling is as follows: First-level scheduling: An appropriate address pool is selected for the domain name based on a load balancing policy. For more information, see Get started. Second-level scheduling: An appropriate address is selected from the address pool. The address pool selects an available address based on a load balancing policy and returns the address to the client. For more information, see Address pool configuration.
Health check template	The health check feature performs real-time probes on the addresses in an address pool to evaluate the operational status and availability of application services. The supported monitoring methods include ICMP ping monitoring, TCP connectivity monitoring, and HTTP/HTTPS endpoint response monitoring. GTM 3.0 provides health check templates that allow you to configure multi-dimensional protocol detection templates to comprehensively determine the availability of application service addresses.	Ping health check TCP health check HTTP(S) health check

How it works

For example, the business domain name of a website is www.example.com.

Add three server IP addresses, 1.1.XX.XX, 2.2.XX.XX, and 3.3.XX.XX, to the GTM address pool and enable health checks.
Configure an access policy as needed.
Activate a GTM instance and complete the basic configuration. A CNAME endpoint is generated based on your configuration. For example: gtm.example.com.
Add a CNAME record to map the business domain name www.example.com to gtm.example.com.

Flowchart

Flow description

An end user sends a DNS query for the domain name of the application service, www.example.com, to a local recursive DNS server.
If the local recursive DNS server does not have a cached entry for www.example.com, it sends a DNS query for the domain name to a root DNS server. The root DNS server then returns the address of the authoritative DNS server for the .com top-level domain (TLD).
The local recursive DNS server receives the address of the authoritative DNS server for the .com TLD from the root DNS server. It then sends a query for www.example.com to the authoritative DNS server for the .com TLD. This server receives the query and returns the address of the authoritative DNS server for example.com to the local recursive DNS server. If the domain name uses Alibaba Cloud DNS, the authoritative server for example.com is an Alibaba Cloud DNS server.
After receiving the address of the Alibaba Cloud DNS server from the authoritative DNS server for the .com TLD, the local recursive DNS server queries the Alibaba Cloud DNS server for www.example.com. In response, Alibaba Cloud DNS finds that www.example.com is an alias for the GTM endpoint gtm.example.com due to a CNAME record, and returns gtm.example.com to the local recursive DNS server.
The local recursive DNS server receives gtm.example.com from the Alibaba Cloud DNS server and then sends a query for gtm.example.com to the GTM DNS server. GTM then uses its pre-configured policies to return the IP address of the application service to the local recursive DNS server.
The local recursive DNS server uses the IP address from the last query as the final resolution for www.example.com. It returns the IP address to the end user and caches it. This allows the local recursive DNS server to directly return the cached result for subsequent queries.
The end user's client receives the IP address from the local recursive DNS server and initiates a direct network connection to the application service.

Service architecture

Description of the service architecture diagram:

The DNS module in the GTM system resolves access requests from end users to the addresses in the application service address pools. For example, users in the Chinese mainland access application services in address pool A, and users outside the Chinese mainland access application services in address pool B. The load balancing policy for both address pools is set to Sequential.
The HealthCheck module in the GTM system performs health checks from multiple regions on the application service addresses in the address pools. These health checks use the ping, TCP, or HTTP(S) method.
If an application service address in address pool A fails a health check, the HealthCheck module detects the failure and notifies the DNS module. The DNS module then temporarily removes the faulty address from the list of available application service addresses. If the HealthCheck module detects that the application service address is available again, the DNS module restores the address to the list of available application service addresses.

This process ensures that end users are automatically routed to the optimal application service by the GTM system, which ensures uninterrupted user access.

System architecture

Global Traffic Manager consists of a control layer and a resolution layer:

Control layer: The control layer provides services through the console and OpenAPI. It is used to perform create, retrieve, update, and delete (CRUD) operations and store data, such as DNS resolution records, configurations, monitoring data, and logs. The control layer is located in the China (Zhangjiakou) and China (Hangzhou) regions.
Resolution layer: The resolution layer provides services through globally deployed resolution server clusters. The resolution layer receives DNS record data from the control layer and responds to DNS queries. The resolution layer has nodes deployed on major continents and in key regions across the globe.

Contact us

DingTalk group: 79530043379