Adaptive Network Telemetry Sampling via Stream Processing

The patent applies dynamics by transitioning from static uniform sampling to dynamic non-uniform sampling. The sampling rate is dynamically adjusted based on network conditions and event priority, allowing the system to collect data more frequently during critical periods and less frequently during normal operation, thereby resolving the contradiction between implementation simplicity and event detection reliability

The patent changes the sampling rate parameter from a fixed uniform value to a variable non-uniform value. By modifying the sampling rate parameter based on detected event types and network conditions, the system achieves both ease of implementation through a straightforward sampling framework and improved reliability through adaptive event detection

The patent applies local quality by differentiating sampling rates for different types of network events. Instead of using a uniform high sampling rate for all events, the system uses higher sampling rates only for critical events (such as security incidents or outages) and lower sampling rates for routine events, thereby improving event detection accuracy for important events while reducing overall resource consumption

The patent applies partial action by selectively applying high sampling rates only when and where needed. Rather than continuously using maximum sampling rate, the system uses partial sampling (higher rate) only for critical event detection, and standard or reduced sampling for non-critical events, achieving adequate measurement precision for important events while avoiding resource exhaustion