Balancing across ActiveGates SaaS ( Modules and synthetic )

[Carl_Me]

Hi, I have a question regarding how Dynatrace manages load balancing across ActiveGates in the following two cases:

1. Multiple ActiveGates with the same module (e.g., Azure or VMware)
2. Multiple ActiveGates in the same Private Synthetic Location

Let's look at a few examples

1. I have two ActiveGates that include the "VMware" module. How does Dynatrace decide which ActiveGate will connect to VMware?
2. When multiple synthetic monitoring systems use a "private synthetic location" containing two ActiveGates, how does Dynatrace balance the load?

My questions mainly concern the initial method of load balancing across the active gates, and whether this allocation becomes static or not. In short, what is the level of load balancing for these two situations?

Spoiler

Obviously, I'm excluding the load balancing process that occurs when an active gate fails ( or other raisons ). This is perfectly normal in that situation.

[t_pawlak · Accepted solution]

Hi,
I will try answer 😉
Let's start with First question:

Multiple ActiveGates with the same module (VMware / Azure)

In these cases you’re typically dealing with remote monitoring / extensions executed on ActiveGate. The key control point is that you select an ActiveGate group for the monitoring configuration, and Dynatrace uses that group to decide where the monitoring runs.
ActiveGate group 
and here u can read about extensions:
About Extensions 

Load balancing behavior (initial + ongoing):

Work is assigned per monitored endpoint (for example, per vCenter / per target), not per individual poll/execution—so an endpoint typically “sticks” to one ActiveGate.

Initially, the scheduler tries to spread endpoints evenly across ActiveGates in the group.

This is largely static rather than real-time CPU/RAM-aware balancing: Dynatrace does not actively monitor AG resource utilization to move jobs away from a busy-but-online ActiveGate.

Reassignment/rebalance happens mainly when the group membership/availability changes (AG down/up), at which point endpoints are redistributed and rebalanced.

For VMware remote monitoring, Dynatrace explicitly states that when creating a monitoring configuration you can pick a corresponding ActiveGate group (often mapped to a network segment).
Simmilar topis u can find here:
Extension Load Balancing on Active Gate Groups 

Second question:

Multiple ActiveGates in the same Private Synthetic Location

Officially, a private location is a group of Synthetic-enabled ActiveGates, and the Dynatrace Cluster schedules executions for that private location. 
https://docs.dynatrace.com/docs/observe/digital-experience/synthetic-monitoring/general-information/...

Dynatrace also states that adding more ActiveGates provides failover and load balancing.

https://docs.dynatrace.com/docs/observe/digital-experience/synthetic-monitoring/synthetic-on-grail/s...

How executions are distributed (practical details):
From Dynatrace Synthetic SMEs / dev-team explanations on the Dynatrace Community:

The configured frequency applies per location (e.g., 5 minutes means every 5 minutes in each assigned location).

For Browser monitors, executions are split across ActiveGates based on how busy they are (an execution may run on AG1, next on AG2, etc.).

Historically for HTTP monitors, there could be a static sharding by monitor definition (a given HTTP monitor repeatedly executed by the same AG), which could lead to imbalance.

With newer behavior described by the Synthetic dev team—when HTTP monitors are executed by the Synthetic engine—distribution becomes more dynamic, with nodes requesting as many runs as they can handle at the moment.

Dynatrace documentation also describes capacity usage for private locations in terms of concurrent executions vs. maximum simultaneous executions, aligning with a capacity-based scheduling view.
here simmilar topis u can check:
https://community.dynatrace.com/t5/Synthetic-Monitoring/Private-synthetic-locations-and-test-frequen...