To improve and provide more flexiblity to current DPOD alerts from rule base to machine learning.

APIC administrators, Operations teams, and application teams will benefit from these improvements, as there will be more options to configure alerts, including:

"APIs That Ended in Error Code Range"

• Adding count and threshold monitoring for APIs that DPOD is tracking
• Including error analysis details from apic-log in the alert emails

Instead of requiring users to input queries or parameters within the alert configuration, the system could be improved by providing dropdown options, as not everyone is familiar with DPOD calls.

For example, the variable used to track 500 errors {"frontendHTTPresponseCodes":"500"} could be simplified with options to select 200, 400, 404, 500, 504 errors, etc.

As our organisation has a very large base of APIs, we cannot customise the API monitoring according to individual API behaviour to prevent false positive alerts.

Therefore, we suggest implementing machine learning capabilities to study API trends.

Different APIs operate and behave differently. For example, some APIs consistently generate 10x 500 errors within an hour, while others may regularly produce 30x 500 errors in the same timeframe. Perhaps DPOD could automatically learn these patterns and only generate alerts when there are significant deviations - such as when API A exceeds its typical 10x threshold or API B surpasses its usual 30x baseline. With this implementation, I believe DPOD would be well-positioned to compete with various log aggregator platforms for a unified view, whether on-premises or in the cloud.