How to setup DC and DR in case of Dynatrace SaaS.

[AK]

Hello Folks,

Is my understanding correct w.r.t DC and DR setup for Dynatrace SaaS?

There will be hosts in DC and DR as well (number of hosts might be less in case of DR)

1. For OneAgent -
   We need to install OneAgents on all the hosts present in DC as well as in DR however monitoring on DR hosts will be disabled and will enabled in case of Disaster (manual work)
2. For ActiveGate -
   We need similar number of ActiveGate, configuration and private synthetic AG if present
3. For Synthetic monitoring -
   Need to import all the monitor/script configurations, credential vaults, certificates (if implemented) and need to assign the appropriate private location if present
4.  Need to replicate or import all other configuration like below -
   - Defined application
   - Custom alerting
   - Custom JMX plugin
   - Integration
   - Maintenance window
   - Management zones
   - Process group naming rules
5. Similar integration must be created for third party ticketing tools like ServiceNow.

Regards,

AK

[ChadTurner · Accepted solution]

As a SaaS customer, DR should be quite minimal as Dynatrace has Cloud clusters handling your instances and in the event an issue arises, new clusters will be provisioned and will take over the work load. 

DR Prep I could see more so for a Managed customer, where I would recommend setting up Dynatrace Backups on a daily basis which will export the Dynatrace settings in the event of a DR issue, you could spin up a new instance and reload it with the back up. Also as a Managed customer, it is recommended to have a cluster set up for failover. Having a 3 cluster node allows you to operate while other locations or nodes have an issue like a Natural Disaster. 

We also had a DR set up, you can read more about it here: https://www.dynatrace.com/news/blog/supporting-disaster-recovery-with-dynatrace/ These hosts were always monitored as core business and in the event of a DR, we had replicas that would spin up with the Dynatrace oneagnet and assume the IPs of the hosts so there was no issue with turning on of off monitoring.