Contents
AD Sizing
This section defines the Workbench Anomaly Detection hardware resources required when deploying the Workbench Anomaly Detection (AD) components.
Workbench Anomaly Detection can be deployed as a single-node/host or as a multi-node/host cluster.
The Workbench Anomaly Detection multi-node cluster deployment is available to support high-availability and/or environments that have a high number of hosts and/or low collection frequency.
AD Node/Host - Cores / Memory / Disk
The minimum hardware requirements for each AD Node/Host is:
- 8 CPU Cores
- 8 GB RAM
- 30 GB HD (free)
Required Number of AD Node(s)/Host(s) at each Workbench Data-Center
Workbench currently supports ingesting Metric data from a maximum of 100 Hosts.
Required Number of AD Nodes/Hosts | Number of Hosts sending Metric data to Workbench | Number of Metrics being sent from each Host to Workbench | Frequency of Metrics being sent from each Host to Workbench |
---|---|---|---|
1 | 100 | 30 (default) | 60 (default) |
1 | 100 | 30 | 30 |
2 | 100 | 30 | 10 |
- Anomaly Detection (AD) Nodes/Hosts should be separate to Workbench Nodes/Hosts - do NOT install AD components on the WB Nodes/Hosts
AD High Availability
Deploy 2 or more AD Nodes/Hosts per Data-Center to provide AD High Availability (HA) - i.e. if 1 AD Node/Host is down Metric data will continue to be processed and Workbench Insights will be generated.