question

pramodANCHUPARAYIL-5135 avatar image
0 Votes"
pramodANCHUPARAYIL-5135 asked GitaraniSharmaMSFT-4262 commented

Regional multi-service failover with Azure Traffic Manager

We have a several services (API, Web APP, Config) that make up an application, deployed in multiple regions with failover in an Active/Passive (hot standby) fashion. Each of those services can be consumed independently and share underlying dataset. Any service failure in one region should initiate a failover to the secondary region for all those services.

How can we set up Azure Traffic manager profiles so an outage in any one of those services will fail over all services together the secondary region. Sort of grouping services that must failover together within a region. I don't see a way to accomplish this using nested profiles.

We have a workaround by creating traffic manager profiles for each service with multi-region endpoints, then set up an Azure function to respond to endpoint down alerts from any profiles to flip each profile/service to the secondary region.

Want to know if there is an easier way to do it, without having to route all traffic thru Azure.

Thanks
P

azure-traffic-manager
· 7
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

The Multi-tier web architecture does not address my requirement. I have a Web App tier, Data Services tier and an API tier. All of which are accessible by customers independantly, any outage in any of the tier should trigger a failover for all tiers to ensure a consistent customer experience.

0 Votes 0 ·

Thank you for the update, @pramodANCHUPARAYIL-5135. Let me check internally regarding your scenario and also discuss this with the product group team for more information. Will keep you updated.

0 Votes 0 ·

Hello @pramodANCHUPARAYIL-5135 ,

Below are the inputs from the Product group team:

The key here is designing your health probe endpoint response. In the most simple sense, the health probe endpoint is a web endpoint that when queried and healthy, will reply ‘200 OK’. In your case, you need to take advantage of the error/response mechanism. Your health probe endpoint should reflect the health of all components of your service. In this way, if any of your three tiers is compromised in any region (however you measure that) can be reflected in the http/https error code returned by a query from the ATM health probe service. This will have to be something you develop. It’s essentially another component in your service – one that specifically attests to multi-tier region health.

Looks like there is no in-built way to achieve your scenario. If you have any specific questions, please let me know and I can further discuss them with the Product group team to provide better clarity.

Regards,
Gita

0 Votes 0 ·

Hello @pramodANCHUPARAYIL-5135 ,

I'm following up on my previous comment. Please let us know if you have any specific questions and I can further discuss them with the Product group team to provide better clarity on this scenario.

Regards,
Gita

0 Votes 0 ·
Show more comments

0 Answers