How to ensure high availability on Azure using Azure Traffic Manager

You have been tasked with migrating your applications to the cloud and here are several tasks that need to be done: Replicate application layer, perform database export / import, or otherwise replicate to cloud, and the main one, upgrade the DNS.

Azure Traffic Manager supports this type of solution by facilitating your DNS management. Let us understand this service can help you in your migration strategy

About Azure Traffic Manager

Azure Traffic Manager is a DNS load balancing service and helps with a number of features, including:

  • Increases application availability
  • Application performance improvement
  • Allows maintenance windows without downtime
  • Combine hybrid applications
  • Distribute traffic to complex deployments

Routing Methods

The traffic manager has routing methods to meet the most varied requirements. Among them we can list:

  • Priority
  • Weight / Weighted
  • Performance
  • Geographical
  • Various Values
  • Subnet


In priority routing, we have the presence of a endpoint primary and the possibility of having multiple endpoints failover. Each endpoint must receive a priority defined by a numeric value from 1 to 1000, where & #8220; smaller & #8221; value has the highest priority.

When the endpoint primary is unavailable, the traffic manager directs traffic to the first endpoint failover available.

Traffic Manager does not route traffic, as the endpoint is selected. It just does the name resolution and after the return of the endpoint for reverse DNS, communication is established directly with the endpoint.

Weight / Weighted

In weighted routing, we can define a Weight to endpoint, thus allowing you to gradually distribute traffic between endpoints. Each of them must receive a numerical value from 1 to 1000, where the highest value have the greater weight.

This routing method has real benefits only when its access load is relatively high. If your traffic is low, the weighted profile may not precisely distribute the weight of the queries.


Unlike the other methods presented, performance-based routing works with respect to latency, and does not necessarily follow any geographic or prioritization rules.

In this case, O endpoint that offering lower DNS query latency will be the endpoint selected for consumption.


Geographic routing takes into account the physical location of the end user. That is, its application is distributed around the planet, the Traffic Manager will route each of the end users in their geographies.

Here it is important to note that after the DNS query, the client will communicate directly with the endpoint. , and not through Traffic Manager.

Various values

This routing method takes into account several endpoints being consulted at once.

Among the methods presented, this is the one that offers the highest availability and lowest latency, since all its endpoints will be consulted at the same time.

On the other hand, since there will be a massive consultation each, you can expect a higher cost to maintain this resource. Not to mention the fact that the only endpoints supported are those with IPv4 or IPv6.


The last method we present here is the subnet routing method. In it, it is possible to define, from a network interval, for which endpoint the user will be directed. This method is particularly useful if you are interested in routing in a targeted manner depending on a specific ISP.


If you are interested in learning more about Azure Traffic Manager and how it can help you with your migration strategy, be sure to check out:

Did you like the content? I invite you to take a look on live that will happen day 23/01/2020 at 20:00. There you will see this content with me in practice and have the chance to ask questions live with me!

See more information here

I wait you there!

Carlos Oliveira

Carlos Oliveira, 25, founder of CloudSquad, a Cloud Computing content sharing blog where he brings tips and tricks about Office 365 and Azure, and is a hub for bringing practical solutions to complex problems.

Carlos Oliveira
pt_BR en_US