Can a load balancer be setup between OneAgents and ActiveGates?

[lawrence_cuneaz]

The Why do I need an ActiveGate? documentation says, "With Dynatrace, you don’t need to worry about load balancing. If you install multiple Environment ActiveGates in the same environment, your OneAgents and Environment ActiveGates will configure themselves automatically to achieve optimal load balancing."

The Configure ActiveGate documentation says the following about configuring a dnsEntryPoint, "This entry can be used if the ActiveGate is accessed via, for example, an external IP address or load balancer."

Can a load balancer be setup between OneAgents and ActiveGates?

My team would like to know, what are the pros and cons of setting up a load balancer in front of the ActiveGates? Do the agents always know when an ActiveGate goes down so that monitoring data is never sent to an ActiveGate offline or does the Agent attempt to send this data to the ActiveGate first and then forward the data on to another ActiveGate if the agent does not receive confirmation?

We are setting up a SaaS deployment but this question should be answered for Managed deployments as well to help other teams.

Thanks,

Lawrence

[Radoslaw_Szulgo · Accepted solution]

Can a load balancer be setup between OneAgents and ActiveGates?

Yes. However, there's usually no need.

what are the pros and cons of setting up a load balancer in front of the ActiveGates?

ActiveGates and OneAgents have several mechanism to optimize the flow of traffic and select the right endpoint - they are by default aware of each server and capable of doing switching between them, basically doing certain load-balancing by themselves. Pros might be some kind of SSL termination better than implemented on AG itself. Usually that's not a problem. Con is that you need to maintain up-to-date redirection list of AGs and basically repeat what AG has embedded.

Do the agents always know when an ActiveGate goes down so that monitoring data is never sent to an ActiveGate offline or does the Agent attempt to send this data to the ActiveGate first and then forward the data on to another ActiveGate if the agent does not receive confirmation?

Yes, OneAgent knows if it fails to send a message to AG. It gets the first endpoint on the list and tries to connect. There's also some switching implemented so the traffic is load balanced. It also has a cache so it can retransmit the data to the other AG so there is no data loss.

[fstekelenburg]

@Radoslaw S. could you extend this answer to Agentless RUM (Javascript Agent), Pros/cons of using an LB, or just two (2-n) CAG's for external recorded user sessions, including session replay?