Monitor the following categories : Performance limit, Utilization Limit, Compute Unit of application gateway of waf v2 tier and standard v2 tier.

Question

Monitor the following categories : Performance limit, Utilization Limit, Compute Unit of application gateway of waf v2 tier and standard v2 tier.

Murarisetty Yamuna 145

Currently I need to monitor the app gateway(WAF V2 & Standard V2 skus) based on following categories : Performance limit, Utilization limit, Compute Units using the application gateway Metrics available in Azure Monitor.
Does the instance count impacts the performance or utilization or compute metrics?
Is there any limit on the count of "capacity units" consumed to load balance the traffic? How to identify the maximum count of capacity units and the compute units?
Is there any limit of "Current connections" of the gateway based on the type of SKU?
How to identify the no. of concurrent connections of the gateway?
If any of the metric reaches threshold value need to notify an alert is needed or not?

KapilAnanth-MSFT 49,536 Reputation points Microsoft Employee Moderator

2023-12-19T09:42:29.9133333+00:00

@Murarisetty Yamuna

Could you please provide an update on this post?

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

Thanks,

Kapil

Please don’t forget to close the thread by clicking "Accept the answer" wherever the information provided helps you, as this can be beneficial to other community members.
Murarisetty Yamuna 145 Reputation points

2023-12-19T16:50:16.7+00:00
@KapilAnanth-MSFT , Thanks for the detailed answer for the requested questions.

As per the document Monitoring Azure Application Gateway and https://learn.microsoft.com/en-us/azure/application-gateway/monitor-application-gateway-reference, I will be categorizing the metrics based on compute units, Performance unit and Utilization.

I have couple of questions which I am mentioning below this mainly concern what metrics in monitoring perspective will belong to performance unit, compute unit and Utilization unit . So these are my assumptions.

To monitor the compute units of the gateway, the key metrics are Current Compute Unit, Current Capacity Unit, Persistent Connections. Is it right?

To monitor the performance of the gateway, key metrics are Unhealthy Host Count, Total Requests, Avg Unhealthy Host Count By BackendPool HttpSettings ? If backend health is unhealthy will it impacts the performance of gateway?

To monitor the utilization unit, the metrics are throughput?
KapilAnanth-MSFT 49,536 Reputation points Microsoft Employee Moderator

2023-12-20T16:49:35.11+00:00
@Murarisetty Yamuna

I am glad we could be of assistance.

1.To monitor the compute units of the gateway, the key metrics are Current Compute Unit, Current Capacity Unit, Persistent Connections. Is it right?

Yes, this is correct

2.To monitor the performance of the gateway, key metrics are Unhealthy Host Count, Total Requests, Avg Unhealthy Host Count By BackendPool HttpSettings ?

If backend health is unhealthy will it impacts the performance of gateway?

"Total Requests" will have an impact on the performance in case you do not have sufficient instances to support the total requests.

As a metric, this is mainly to have an overview of the total expected connections at any time.

The other 2 metrics are to indicate the "Health" of the backend.

If the backend is unhealthy, the requests will simply fail as there is no one to serve the requests. There is no performance to monitor when "all" the traffic is going to fail.

i.e., It can be considered an indicator in a binary format - your backend is either up or not

3.To monitor the utilization unit, the metrics are throughput?

This is quite incorrect.

For Utilization, the metrics are "Current capacity units" and "Current compute units".

Check this :

Cheers,

Kapil
Murarisetty Yamuna 145 Reputation points

2024-01-08T13:15:34.95+00:00
@KapilAnanth-MSFT ,

Based on the information provided, if the current compute units and current capacity units exceed 75%, it indicates maximum usage. An alert should be set up to notify when this threshold is crossed. The challenge is how to map this percentage to count units in order to specify the threshold value for the metric signals of compute units and capacity units?

The average current capacity unit and current compute unit over the last 30 days are as follows:

Average Current Capacity Unit: 0.80 units

Average Current Compute Units: 0.80 units
KapilAnanth-MSFT 49,536 Reputation points Microsoft Employee Moderator

2024-01-15T08:11:36.09+00:00
@Murarisetty Yamuna

The alert logic here states that you must monitor for 75% of your use case/workload.

So, if your case is

Average Current Capacity Unit: 0.80 units

Average Current Compute Units: 0.80 units

Then you should consider 75% of them to configure the alert,

Current Capacity Unit : 0.60 (75% of .80 units)

Current Compute Unit : 0.60 (75% of .80 units)

P.S : Please note the document was stating Peak usage for Capacity Unit and not Average (i.e., Maximum)

Cheers,

Kapil

Accepted answer

0 additional answers

Your answer

KapilAnanth-MSFT 49,536 Reputation points Microsoft Employee Moderator

2023-12-19T09:42:29.9133333+00:00

@Murarisetty Yamuna

Could you please provide an update on this post?

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

Thanks,

Kapil

Please don’t forget to close the thread by clicking "Accept the answer" wherever the information provided helps you, as this can be beneficial to other community members.
Murarisetty Yamuna 145 Reputation points

2023-12-19T16:50:16.7+00:00

@KapilAnanth-MSFT , Thanks for the detailed answer for the requested questions.

As per the document Monitoring Azure Application Gateway and https://learn.microsoft.com/en-us/azure/application-gateway/monitor-application-gateway-reference, I will be categorizing the metrics based on compute units, Performance unit and Utilization.

I have couple of questions which I am mentioning below this mainly concern what metrics in monitoring perspective will belong to performance unit, compute unit and Utilization unit . So these are my assumptions.

To monitor the compute units of the gateway, the key metrics are Current Compute Unit, Current Capacity Unit, Persistent Connections. Is it right?

To monitor the performance of the gateway, key metrics are Unhealthy Host Count, Total Requests, Avg Unhealthy Host Count By BackendPool HttpSettings ? If backend health is unhealthy will it impacts the performance of gateway?

To monitor the utilization unit, the metrics are throughput?
KapilAnanth-MSFT 49,536 Reputation points Microsoft Employee Moderator

2023-12-20T16:49:35.11+00:00

@Murarisetty Yamuna

I am glad we could be of assistance.

1.To monitor the compute units of the gateway, the key metrics are Current Compute Unit, Current Capacity Unit, Persistent Connections. Is it right?

Yes, this is correct

2.To monitor the performance of the gateway, key metrics are Unhealthy Host Count, Total Requests, Avg Unhealthy Host Count By BackendPool HttpSettings ?

If backend health is unhealthy will it impacts the performance of gateway?

"Total Requests" will have an impact on the performance in case you do not have sufficient instances to support the total requests.

As a metric, this is mainly to have an overview of the total expected connections at any time.

The other 2 metrics are to indicate the "Health" of the backend.

If the backend is unhealthy, the requests will simply fail as there is no one to serve the requests. There is no performance to monitor when "all" the traffic is going to fail.

i.e., It can be considered an indicator in a binary format - your backend is either up or not

3.To monitor the utilization unit, the metrics are throughput?

This is quite incorrect.

For Utilization, the metrics are "Current capacity units" and "Current compute units".

Check this :

Cheers,

Kapil
Murarisetty Yamuna 145 Reputation points

2024-01-08T13:15:34.95+00:00

@KapilAnanth-MSFT ,

Based on the information provided, if the current compute units and current capacity units exceed 75%, it indicates maximum usage. An alert should be set up to notify when this threshold is crossed. The challenge is how to map this percentage to count units in order to specify the threshold value for the metric signals of compute units and capacity units?

The average current capacity unit and current compute unit over the last 30 days are as follows:

Average Current Capacity Unit: 0.80 units

Average Current Compute Units: 0.80 units
KapilAnanth-MSFT 49,536 Reputation points Microsoft Employee Moderator

2024-01-15T08:11:36.09+00:00

@Murarisetty Yamuna

The alert logic here states that you must monitor for 75% of your use case/workload.

So, if your case is

Average Current Capacity Unit: 0.80 units

Average Current Compute Units: 0.80 units

Then you should consider 75% of them to configure the alert,

Current Capacity Unit : 0.60 (75% of .80 units)

Current Compute Unit : 0.60 (75% of .80 units)

P.S : Please note the document was stating Peak usage for Capacity Unit and not Average (i.e., Maximum)

Cheers,

Kapil

Answer 1

@Murarisetty Yamuna

Welcome to the Microsoft Q&A Platform. Thank you for reaching out & I hope you are doing well.

Wrt your queries,

1.Currently I need to monitor the app gateway(WAF V2 & Standard V2 skus) based on following categories : Performance limit, Utilization limit, Compute Units using the application gateway Metrics available in Azure Monitor.

2.Does the instance count impacts the performance or utilization or compute metrics?

Instance count impacts performance in case there is a load on the App gateway and the number of instances are unable to keep up with the Load.
Refer to "Autoscaling" in Autoscaling and High Availability
You can refer to this document to understand how to Tune the Application gateway upper and lower instance count limit based on your application's load
NOTE : The instance count does not affect metric collection.

3.Is there any limit on the count of "capacity units" consumed to load balance the traffic? How to identify the maximum count of capacity units and the compute units?

Capacity units and Compute Units are not something the customer is expected to configure and worry about
Your focus should be on "Instance Count"
Details on how the Instance Count is related to Capacity unit and Compute Unit is documented here.
With that said, the limit on the maximum Instance Count is defined as 125.
Theoretical max capacity units : Each instance guarantees a minimum of 10 capacity units in terms of processing capability. So the max would be 125*10 = 1250 units.

4.Is there any limit of "Current connections" of the gateway based on the type of SKU?

Again, the max Instance count is 125. Each instance guarantees a minimum of 10 capacity units in terms of processing capability. A Capacity Unit can support 2500 Persistent connections.
So, theoretically, the max would be 125102500 Persistent connections.
I couldn't find the upper limit for v1 SKU. If you are interested with v1 SKU, I can check internally.
However, you must note that you are charged for "Capacity Unit" of v2 SKU.

5.How to identify the no. of concurrent connections of the gateway?

You can use the metric "Current connections" from Application Gateway metrics

6.If any of the metric reaches threshold value need to notify an alert is needed or not?

Various use cases are covered in Alerts for Application Gateway v2 SKU (Standard_v2/WAF_v2).
Following the metrics listed here should do the trick.

Thanks,

Kapil

Please don’t forget to close the thread by clicking "Accept the answer" wherever the information provided helps you, as this can be beneficial to other community members.

Share via

Monitor the following categories : Performance limit, Utilization Limit, Compute Unit of application gateway of waf v2 tier and standard v2 tier.

0 additional answers

Your answer