Introduction to CA Anomaly Detector › Data Collection › Scalability

Scalability

CA Anomaly Detector is highly scalable. The program does not retain a lot of data. It collects and analyzes only the results of data analysis from the data sources that it is configured to analyze.

The MySQL database does not become overloaded easily. For example, after several weeks of running CA Anomaly Detector at a large enterprise, the database growth rate was equivalent to 4 GB per year. The typical growth rate in testing is 160,000 database rows per day.

Although you can run CA Anomaly Detector with most Harvesters, some limitations apply. In our testing, the following configurations were the maximum configurations that had acceptable performance:

• Linux-based Harvester that runs a maximum of 2 million flows per minute
• Windows-based Harvester that runs a maximum of 4 million flows per minute and does not run the Flow Cloner
• Windows-based Harvester that runs a maximum of 2 million flows per minute and runs the Flow Cloner

We support using a single instance of CA Anomaly Detector to monitor the following data sources:

• CA Network Flow Analysis: Single instance (Required)
• CA NetVoyant: Two instances (Optional)
• CA Application Delivery Analysis (SuperAgent): Two instances (Optional)
• CA Unified Communications Monitor: Two instances (Optional)