@승필 박 Welcome to the Microsoft Q&A Platform. Thank you for reaching out & I hope you are doing well. For a Load Balancer,
- Azure Load Balancer is built to handle millions of requests per second while ensuring your solution is highly available
- While we do not have definite numbers that we can share, it is safe to say that LB will support as many connections as the backend VMs could handle.
-
- Refer : Azure Load Balancer
For an Application Gateway V2,
- The calculation is based on Capacity Unit
- There are three dimensions to capacity unit - compute unit, persistent connections, and throughput
- Each capacity unit is composed of at most: 1 compute unit, 2500 persistent connections, and 2.22-Mbps throughput.
- The above mentioned metrics are abstracted (i.e, not directly configurable) from Portal
- What you can configure is a Instance Count. Each instance guarantees a minimum of 10 capacity units in terms of processing capability. The same instance could potentially support more than 10 capacity units for different traffic patterns depending upon the Capacity Unit parameters.
- You can find more details at Capacity Unit , Compute Unit, Instance Count
- So, the performance depends upon the no. of instances deployed into the AppGateway.
Since each Instance is guaranteed to have 10 capacity units, we can say that for an App gateway with only one instance, we can have 25000 persistent connections, and 22.2-Mbps throughput. This can be improved by scaling the App gateway instances. 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.