Microburst Detection via Aggregated Data to Reduce Network Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting microburst traffic, such as capturing packets on a microsecond basis and aggregating them on a millisecond or microsecond basis, result in heavy network load between collection and monitored devices, and acquiring device states on a millisecond or microsecond basis also poses a heavy load on monitored devices.
Innovation Solution
An information processing apparatus that receives aggregate-value information of data communication and accumulation amounts aggregated over one second or longer, and analyzes these values to determine if microburst traffic has occurred, reducing the load required for detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If packets are captured on a microsecond basis and aggregated on a millisecond/microsecond basis, then microburst detection capability is improved, but network load between collection device and monitored device becomes heavy
Solution Approach 1:
The monitored device performs preliminary aggregation of packet data at its own location before transmitting to the collection device. By pre-aggregating data on a millisecond basis at the monitored device, the system reduces the amount of data transmitted over the network, thereby reducing network load while maintaining microburst detection capability.
Solution Approach 2:
The monitored device acts as an intermediary that performs local aggregation before forwarding data to the collection device. This intermediate aggregation step reduces the volume of data transmitted across the network, reducing network load while preserving the ability to detect microbursts through the aggregated data.
2Measurement precision
If the state of a monitored device is acquired on a millisecond/microsecond basis, then microburst detection capability is improved, but load on the monitored device becomes heavy
Solution Approach 1:
The monitored device performs self-service by aggregating its own packet data locally before transmitting to the collection device. This self-aggregation at the monitored device reduces the processing load on the device compared to continuous high-frequency state acquisition, while still enabling microburst detection through the aggregated data.
3Loss of energy
If aggregate-value information is aggregated per unit of aggregation time of one second or longer, then network load and device load are reduced, but detection speed may be affected
Solution Approach 1:
The system changes the aggregation time parameter to one second or longer, which reduces the frequency of data collection and processing. This parameter change lowers network load and device load significantly, while the aggregation process itself maintains detection accuracy by summarizing data over appropriate time windows that capture microburst characteristics.
Data Source
AI summary
A notification acquiring section receives, from a device, a notification message indicating at least one of an aggregate value of data communication amounts obtained by aggregating, per unit of aggregation time of one second or longer, the data communication amounts of the device which have been measured on a millisecond or microsecond basis; and an aggregate value of data accumulation amounts obtained by aggregating, per unit of the aggregation time, the data accumulation amounts in the device which have been measured on the millisecond or microsecond basis. A determining section analyzes at least one of the aggregate values of the data communication amounts and the data accumulation amounts indicated by the notification message to determine whether or not microburst traffic has occurred at the device.