Dynatrace Managed rack awareness groups Cluster nodes into three fault domains (racks). A rack-aware deployment is resilient to an outage of all nodes in a rack. You can make a single Managed Cluster rack-aware, or apply rack awareness to each data center in a Premium High Availability (PHA) deployment.
Rack awareness ensures that no replica is stored redundantly inside a single rack, so replicas are spread across all racks. If one rack goes down, the other two full replicas remain available, which keeps data consistent and accessible. For example, in the deployment below, the Managed Cluster survives the loss of all three nodes in Rack 1 without data loss.
| Data center | Rack | Nodes | Status |
|---|---|---|---|
Data center 1 | Rack 1 | Node 1, Node 2, Node 3 | Down |
Data center 1 | Rack 2 | Node 4, Node 5, Node 6 | Up |
Data center 1 | Rack 3 | Node 7, Node 8, Node 9 | Up |
A standard Dynatrace Managed high-availability deployment needs at least three Cluster nodes to prevent data loss. A rack-aware deployment needs three racks (fault domains) for the same reason: if one rack fails, the surviving two racks still hold a complete copy of the data. Because each rack holds at least three nodes, a rack-aware deployment tolerates the loss of an entire rack.
The same concept applies to PHA deployments. Rack-aware Managed Clusters in separate data centers increase your resilience to data loss.
Without rack awareness, the nodes in a data center aren't grouped into fault domains. In the following example, Data center 1 is offline and one node in Data center 2 is also down. The remaining eight nodes in Data center 2 still hold a full replica set, so no data is lost.
| Data center | Nodes | Status |
|---|---|---|
Data center 1 | Node 1 to Node 9 | Down |
Data center 2 | Node 1 | Down |
Data center 2 | Node 2 to Node 9 | Up |
With rack awareness, the same PHA deployment tolerates the loss of a whole rack in the surviving data center. In the following example, Data center 1 is offline and Rack 1 in Data center 2 also fails. The two remaining racks in Data center 2 each hold a complete replica, so no data is lost.
| Data center | Rack | Nodes | Status |
|---|---|---|---|
Data center 1 | Rack 1 | Node 1, Node 2, Node 3 | Down |
Data center 1 | Rack 2 | Node 4, Node 5, Node 6 | Down |
Data center 1 | Rack 3 | Node 7, Node 8, Node 9 | Down |
Data center 2 | Rack 1 | Node 1, Node 2, Node 3 | Down |
Data center 2 | Rack 2 | Node 4, Node 5, Node 6 | Up |
Data center 2 | Rack 3 | Node 7, Node 8, Node 9 | Up |
For the highest resilience, make every data center in a PHA deployment rack-aware.
Use rack awareness only if:
Otherwise, you might lose data and encounter Managed Cluster availability issues.
A deployment becomes rack-aware either at installation time or through a conversion. During the initial Managed installation, the --rack-dc and --rack-name parameters assign each node to a data center and a rack. For the parameter descriptions, go to Customize installation for Dynatrace Managed.
An existing Managed Cluster converts to rack-aware in one of two ways, and the size of your metric storage decides which one fits:
To run rack awareness together with Premium High Availability, go to Combine Premium High Availability with rack awareness.