Global Traffic Manager
Global Traffic ManagerConcept-related questions
Global Traffic Manager (GTM) vs. Server Load Balancer (SLB)
GTM uses DNS to resolve domain names to multiple IP addresses, distributing application traffic by directing users to different IP addresses. GTM also dynamically updates the list of IP addresses that DNS returns by performing health checks, enabling fault isolation and failover. End users connect directly to the service’s IP address and do not route traffic through GTM. In contrast, SLB proxies user requests and distributes them in real time to different servers, so end-user traffic must pass through SLB.
Use SLB for load balancing within the same region. Use GTM to balance traffic across multiple SLB endpoints in different regions.
The following table compares the two:
Comparison item | Network layer | Backend address | Weighted round-robin | Cross-region complexity | Fault isolation time | Session persistence |
Global Traffic Manager | Layer 3 | Domain name or IP | Supported | Simple | minute-level | Not supported |
Server Load Balancer (SLB) | Layer 4 and Layer 7 | IP | Supported | Difficulty | second-level | Supported |
Global Traffic Manager (GTM) vs. Alibaba Cloud DNS
GTM builds on intelligent DNS resolution. It adds health checks and automatic failover. GTM routes users to the nearest endpoint based on their geographic location. It monitors service status in real time.
Domain access questions
Can multiple business domains share one GTM instance?
It depends.
If all your business domains resolve to the exact same set of IP addresses, use CNAME records to point them to the same GTM access domain. Otherwise, each business domain needs its own GTM instance.
1. Using one GTM instance
The business domain www.example.com resolves to 1.1.XX.XX and 2.2.XX.XX. These two IP addresses require disaster recovery.
The business domain test.example.com also resolves to 1.1.XX.XX and 2.2.XX.XX. These two IP addresses also require disaster recovery.
In this case, both domains resolve to the same IP address set. You need only one GTM instance. Set a CNAME record for www.example.com and another for test.example.com. Point both records to the GTM access domain. For details, see 1. Configure the access domain.
2. Scenario where you cannot use the same GTM instance
The business domain www.example.com resolves to 1.1.XX.XX and 2.2.XX.XX. These two IP addresses require disaster recovery.
The business domain test.example.com resolves to 1.1.XX.XX and 3.3.XX.XX. These two IP addresses also require disaster recovery.
In this case, the domains resolve to different IP address sets. Each domain needs its own GTM instance.
Global Traffic Manager (GTM) access domain: can I access it directly by URL?
Yes. The GTM access domain is a valid, directly accessible URL. You can also use it as the CNAME target for other business domains that serve end users.
Global Traffic Manager (GTM): how does GTM detect service failure?
GTM includes built-in service monitoring. It uses global monitoring nodes. GTM triggers alerts when multiple monitoring points report failures. You can choose Ping, TCP, or HTTP(S) health checks to monitor your service.
Ping health check: detects failure based on packet loss rate and response time.
TCP health check: detects failure based on port response time.
HTTP/HTTPS health check: detects failure based on response time and return code.
Global Traffic Manager (GTM) failover: how long does it take to take effect?
Answer: Extensive testing shows that when an application service fails, the GTM Ultimate Edition can detect the fault and initiate a failover in about 1 minute. The global fault recovery time is the sum of the fault detection time and the global effective time.
Standard Edition: failure detection and failover in about three minutes
Failure detection time: With a health check interval of 60 seconds, TTL of 60 seconds, and two consecutive failures, GTM detects and fails over in about three minutes.
Network-wide propagation time: GTM cannot guarantee a fixed propagation time. Propagation depends on ISP TTL cache settings and network conditions across regions.
Ultimate Edition: failure detection and failover in about one minute
Failure detection time: With a health check interval of 15 seconds, TTL of 1 second, and three consecutive failures, GTM detects and fails over in about one minute.
Network-wide propagation time: GTM cannot guarantee a fixed propagation time. Propagation depends on ISP TTL cache settings and network conditions across regions.
Global Traffic Manager (GTM) address pool: can I use domain names?
Yes. An address pool can contain IP addresses or domain names. Do not mix IP addresses and domain names in the same pool. If you add multiple domain names, GTM performs round-robin DNS resolution by default.
Global Traffic Manager (GTM): does GTM support intelligent DNS resolution?
Yes. GTM includes intelligent DNS resolution. It supports routing based on the user’s ISP, region, or country. GTM currently supports China ISPs, seven regions in China, six continents outside China, and selected countries. This ensures users connect to the nearest endpoint and improves access speed.
Global Traffic Manager (GTM): does GTM support session persistence?
No. GTM operates at the DNS level. It routes clients to application endpoints using DNS responses. Clients connect directly to the application IP address. GTM does not see HTTP traffic between clients and servers. So GTM cannot support session persistence.
Can a domain name use Global Traffic Manager (GTM) and CDN at the same time?
Yes. Place CDN in front of GTM. For details, see CDN > M integration for secure access and high availability.
Global Traffic Manager (GTM) address pool: can I add a CDN provider’s CNAME access domain?
You can add it, but we do not recommend it. CDN has many nodes. GTM has limited health check nodes. This may cause inaccurate monitoring and failover, reducing health check and failover reliability.
DNS resolution does not take effect. Returns old IP or NXDOMAIN.
A: It takes time for Global Traffic Manager DNS records to take effect. Please wait patiently. If the records still do not take effect after an extended period, refer to the following steps to troubleshoot the issue:
Check if TTL expired: run
dig www.example.comto view the TTL countdown.Check CNAME configuration: run
dig +trace www.example.com.Clear local DNS cache: run
sudo systemctl flush-dns.
Health check fails
If health checks fail but your service is running, follow these steps:
Confirm your firewall allows traffic from GTM probe IP ranges.
Test the health check path: run
curl -H "Host: domain" http://ip:port/path.Check server load and response time.
Access domain record type is A record. Address pool contains domain names. Why does dig on the access domain return no CNAME result?
Based on the constraints in Configure Address Pools, A-type access domain names support the following address pool types: A and Domain Names. However, they respond only to A-type requests. To respond to CNAME-type requests, you must change the access type of the Access Domain to CNAME type.
Billing questions
One address is used by both subscription and pay-as-you-go instances. How are health checks billed?
Each instance type is billed separately. Subscription instances are billed per probe task triggered by the access domain. Pay-as-you-go instances are billed per health check probe.
Alert questions
Why did I not receive an alert notification?
If you do not receive expected alerts, follow these steps:
Check alert rules: Confirm the rule exists and is enabled.
Check contact status: On the alert contact management page, confirm the phone number or email is verified.
Check spam or blocked messages: Check your email spam folder or SMS block list.
Check health check logs: Confirm the monitored address or address pool changed state as defined by the alert rule.
What triggers an alert rule?
Alert rules trigger on health check status changes. Common triggering events include the following:
Address unavailable: Triggered when an address fails health checks consecutively for the configured threshold.
Address recovered: Triggered when an unavailable address passes health checks again.
Address pool unavailable: Triggered when all addresses in the pool become unavailable.
Address pool recovered: Triggered when at least one address in an unavailable pool becomes available again.