Application Gateway – Review / Clarification on Current Traffic Architecture and Behavior

Question

Application Gateway – Review / Clarification on Current Traffic Architecture and Behavior

Kanifnath Deshmukh (Infosys Ltd) 0 Microsoft External Staff

Hello Team,

Currently, our traffic is routed through a backdoor configuration where requests flow directly from Azure Traffic Manager to Azure App Service, without passing through Application Gateway.

We are in the process of transitioning to a new architecture where traffic flows as follows:

Traffic Manager → Application Gateway → App Service

In production, we have 7 regions. We began gradually migrating region by region by closing the backdoor route and directing traffic through Application Gateway.

However, after enabling traffic through Application Gateway in one of the regions, we observed a significant increase in latency reported by our partner systems that consume our service. The latency increase was noticeable and impacted their calls. Please refer to the attached screenshot for details.

Due to the impact, we had to roll back the change and restore traffic to the previous backdoor routing (Traffic Manager → App Service).

We would like assistance in understanding:

Why the latency increased significantly after introducing Application Gateway.

Whether there are any configuration aspects (listeners, HTTP settings, probes, WAF, autoscaling, etc.) that could contribute to this behavior.

If any known performance considerations should be reviewed when placing Application Gateway in the request path.

Kindly help us analyze the possible root cause of this latency increase.

Thank you.

We currently support seven regions, with the four regions listed below accounting for approximately 80% of overall traffic.

Application gateway :

agw-pims-prod-cus-1

agw-pims-prod-cus-2

agw-pims-prod-ncus-1

agw-pims-prod-ncus-2

Venkatesan S 4,825 Reputation points Microsoft External Staff Moderator

2026-03-03T23:20:07.32+00:00
Hi Kanifnath Deshmukh (Infosys Ltd),

Thanks for reaching out in Microsoft Q&A forum,

Why the latency increased significantly after introducing Application Gateway. Whether there are any configuration aspects (listeners, HTTP settings, probes, WAF, autoscaling, etc.) that could contribute to this behavior.

Azure Application Gateway between Traffic Manager and App Service naturally adds latency through extra processing layers like SSL termination, routing, and backend checks especially noticeable in your high-traffic cus-1/2 and ncus-1/2 regions handling 80% of volume.

Gateway operations such as URL routing, session affinity, and health probes add milliseconds per request versus direct App Service access. WAF (likely enabled on agw-pims-prod-*), inspects/buffers every payload, often doubling latency during spikes. Autoscaling v2 reacts in 3-5 minutes if min instances are too low for baseline load.

Check and tune these areas on your gateways:

WAF Policy: Full inspection overhead switch to Detection mode; exclude noisy rules to baseline performance.

HTTP Backend Settings: Short timeouts hurt flow increase request timeout to 60s max; enable HTTP/2 and compression.

Health Probes: Frequent polling strains backends use 30s interval, 30s timeout, threshold 3 with custom /health path.

Autoscaling: Low min capacity causes ramp-up delays set based on peak Capacity Units (5-10+ minimum for your scale).

Backend Pool: Public IPs risk SNAT exhaustion switch to VNet-integrated private endpoints for App Service.

Monitor Backend Response Time vs Total Gateway Time metrics first. Test one region with WAF off/tuned scaling before rollout. Enable zone redundancy, connection draining, and /24+ subnet sizing.

Reference:

Architecture Best Practices for Azure Application Gateway v2 - Microsoft Azure Well-Architected Framework | Microsoft Learn

Use Azure App Gateway with Azure Traffic Manager | Microsoft Learn

Azure Application Gateway Backend Settings configuration | Microsoft Learn

Monitor Azure Application Gateway | Microsoft Learn

Kindly let us know if the above helps or you need further assistance on this issue.

Please “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
NITIN GUPTA 1 Reputation point

2026-03-04T00:05:44.4966667+00:00
Hi Kanifnath,

Thank you for posting in Microsoft Q&A.

To assist you better, could you please provide the following information:

In which region has the Application Gateway been deployed?

Is the application deployed across seven regions with different URLs for each region, or is a single URL being used for all regions?

Regards,

Nitin
Kanifnath Deshmukh (Infosys Ltd) 0 Reputation points Microsoft External Staff

2026-03-06T18:49:43.6933333+00:00

Hi Nitin,

Application gateway is deployed in East us2, west us 3, North central us and central us. For each app service instance we have separate application gateway

1 answer

Your answer

NITIN GUPTA 1 Reputation point

2026-03-04T00:05:44.4966667+00:00

Hi Kanifnath,

Thank you for posting in Microsoft Q&A.

To assist you better, could you please provide the following information:

In which region has the Application Gateway been deployed?

Is the application deployed across seven regions with different URLs for each region, or is a single URL being used for all regions?

Regards,

Nitin
Kanifnath Deshmukh (Infosys Ltd) 0 Reputation points Microsoft External Staff

2026-03-06T18:49:43.6933333+00:00

Hi Nitin,

Application gateway is deployed in East us2, west us 3, North central us and central us. For each app service instance we have separate application gateway

Answer 1

Hello Kanifnath Deshmukh (Infosys Ltd)

**
**Please review the application gateway access logs to analyze the time taken and server latency values and determine how long the server takes to respond to the application gateway.

If the server's response time is equal to or greater than the application gateway, then the latency is not due to the application gateway, which is expected. The application gateway acts as a load balancer and will reflect the response time of the backend server. If you notice higher server latency, please address the issue on the server side first.

If you find any latency in the application gateway logs, let me know and I can assist further.

Check the reference document to check the logs:
https://learn.microsoft.com/en-us/azure/application-gateway/monitor-application-gateway

And also check the document Use Azure App Gateway with Azure Traffic Manager

Similarly, for the traffic manager, you can review the diagnostic logs and check the status (Up or Down) at the time the issue occurred to help resolve the problem.

Additionally, if you're interested, I can recommend using Front Door instead of Traffic Manager. It is also a global load balancer with similar capabilities, and Front Door offers a caching feature as well. If you want to use the cache option, you can choose Front Door. And it will support WAF also.

Check the reference document:
https://learn.microsoft.com/en-us/azure/frontdoor/front-door-overview

I hope the above answer helps you! Please let us know if you have any further questions.

Please don't forget to "accept the answer" and "upvote" where the information provided will help you, this can be beneficial to other members of the community.

Kanifnath Deshmukh (Infosys Ltd) 0 Reputation points Microsoft External Staff

2026-03-06T18:47:43.2233333+00:00

Hi Praveen,

Thanks for sharing all the details. Based on the logs, the overall time taken and the server‑side latency appear to be almost identical. When I look specifically at the latency from the App Service itself, the numbers look healthy. What’s unclear is why the latency increases only when the App Service is invoked through the Application Gateway, rather than when it’s called directly through the Traffic Manager. This discrepancy suggests that something in the App Gateway path may be contributing to the additional delay.
Praveen Bandaru 10,720 Reputation points Microsoft External Staff Moderator

2026-03-16T17:18:26.7466667+00:00

Hello Kanifnath Deshmukh (Infosys Ltd)

This is expected behavior however, with the Application Gateway, each request is directed to a specific backend, and it waits for a response from that backend. In your case, the backend is taking longer to respond, which is causing the latency. We have observed this in your logs as well. The Traffic Manager acts as a global load balancer and delivers responses from the same backend if it is healthy.

Share via

Application Gateway – Review / Clarification on Current Traffic Architecture and Behavior

1 answer

Your answer