How Trafficmanager (traffic manager) is routed in Microsoft Azure

Today's traffic manager for Azure has three alternative routing options. Although you can choose any routing method at any time, the configuration file for each traffic manager can only use one route method at a time period.

It is important to note that all traffic routing methods include endpoint monitoring. After configuring the Traffic Manager profile to specify the traffic routing method that best fits your needs, you need to configure monitoring settings. When monitoring is configured correctly, traffic manager monitors the status of endpoints, including cloud services and websites, and does not send traffic to endpoints it deems to be unavailable.

The three traffic manager traffic routing methods are: (For the sake of understanding, here are the scenarios cited)

1, failover: You have nodes in the same or different azure datacenters and you want to specify that some nodes prioritize all traffic, and once the specified node fails, switch immediately to the node that made the backup. In this scenario, you can choose to fail over.

2, round robin: When you want to distribute load between nodes in the same data center or across different data centers, you can choose round robin.

3, Performance: When your access comes from a different geographic location, you want the client to use "closest" node access to achieve the lowest latency and can choose performance.

It is important to note that the Web site on Azure (website) itself provides the ability to route failover and cyclic traffic within the same datacenter, regardless of the mode of the site. Traffic Manager allows you to specify the failover and looping traffic for your site to be routed on different datacenters.

Methods for routing failover traffic

Typically, an organization is guaranteed to provide the reliability of its services. This is achieved primarily by switching to a backup node immediately when the primary node is down. A typical scenario for failover is to provide a set of identical nodes and send traffic to the master node while sending to one or more backup nodes. If the primary node is unavailable, the client sending the request is re-directed to the other node. If the first standby node is unavailable, it will point to the second, third, until the available nodes are found. When configuring the methods for failover business routes, it is important to select the order of the endpoints. When configuring the failover service route method, the order of the selected endpoints is important.

Round-robin Routing method

A common round-robin routing pattern is to provide a set of identical nodes that send traffic to a node that has been circulating in a way. It randomly selects one of the fastest nodes and does not send traffic to the detected unavailable nodes. Round robin also supports weighted allocation of network traffic.

The weighted distribution of rotations allows you to assign a load to the allocation weights based on each node. The higher the weight value, the more frequently it is accessed. The scenarios available for this method include:

1, gradual application upgrade: Allocate the route percentage of traffic to a new endpoint, and gradually increase the traffic over time to 100% to complete the application upgrade.

2, the application migrates to Azure: Create configuration files that are both Azure and external endpoints, and specify weights to route traffic to each node.

3, explosive growth: quickly deploy to the cloud with traffic manager profiles to cope with explosive growth. When you need additional capacity in the cloud, you can add or enable multiple endpoints and specify which portions of the traffic go to each endpoint.

At this point, you cannot configure weighted traffic routing using Azure's classic portal. Azure provides programmatic access to the method using the related service Management Rest API and Azure PowerShell commands.

If you want to test the profile of a single client and observe whether it equals a weighted round robin behavior, verify that DNS name resolution differs from the IP address of the same or weighted endpoint in the configuration file. When you test, you must first disable the client DNS cache or clear each DNS cache that tries to ensure that a new DNS name query is sent.

Performance routing Methods

To distribute traffic to nodes located in different datacenters around the world, you can import traffic directly to the nearest node to guarantee the minimum latency between the client and the endpoint. Typically, the "closest" endpoint is directly equivalent to the nearest geographic distance. The performance traffic routing method allows you to allocate traffic based on location and latency, but does not take into account the real-time changes in the network configuration or the load account.

The performance routing method is to find and request the node closest to the client. The closest node here is determined by the Internet latency and the round-trip time between each IP address and each data center. It is periodically updated and does not imply a real-time reflection across Internet performance. It does not take into account the load of a given service, even if traffic manager monitors the nodes based on the method you choose and does not include DNS query responses if they are not available. In other words, performance traffic routing also uses the method of failover traffic routing.

This article from the "11142118" blog, reproduced please contact the author!

How Trafficmanager (traffic manager) is routed in Microsoft Azure