Adaptive Encryption Compression for CPU Load Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Network Virtualization Security Platforms, data encryption is resource-intensive, and combining compression with encryption can either reduce or increase CPU load, depending on the data characteristics, making it challenging to predict whether compressing data before encryption will conserve CPU resources.
Innovation Solution
An adaptive approach that predicts whether compressing a data packet before encryption will reduce CPU load by analyzing historical throughput data and entropy-based properties of the packet, using a weighted average ratio of compression to determine if compressing the packet would decrease the CPU load required for encryption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If data compression is applied before encryption, then the size of data to be encrypted is reduced, but the CPU resources required for compression are consumed
Solution Approach 1:
The patent dynamically adjusts the compression threshold parameter based on packet entropy characteristics. By changing the parameter (compression threshold) according to the data's entropy, the system optimizes the balance between compression benefits and CPU resource consumption, applying compression only when it will result in net CPU savings.
Solution Approach 2:
The system implements dynamic decision-making by evaluating packet entropy and comparing it against adaptive thresholds. Rather than statically compressing all or no packets, the system dynamically determines on a per-packet basis whether compression should be applied, allowing flexible adaptation to varying data characteristics and system conditions.
2Productivity
If compression is applied to all packets, then network throughput may increase, but CPU utilization increases due to the additional compression processing
Solution Approach 1:
The patent applies compression selectively to specific packets based on their individual entropy characteristics rather than uniformly to all packets. By identifying and compressing only those packets with low entropy (which compress well), the system achieves throughput improvement in specific cases without incurring CPU overhead for packets where compression would not be beneficial.
Solution Approach 2:
The system uses entropy-based parameter adjustment to determine compression eligibility. By changing the decision parameter (whether to compress) based on measured entropy values, the system optimizes the balance between throughput gains and CPU utilization, ensuring compression is applied only when it provides net benefit.
3Quantity of substance
If compression is applied to high-entropy data, then CPU resources are consumed without significant size reduction
Solution Approach 1:
The patent fundamentally changes the approach by using entropy as a parameter to gate the compression decision. By measuring packet entropy and comparing it to a threshold, the system identifies which packets are suitable for compression (low entropy) and which are not (high entropy), avoiding wasted CPU resources on packets where compression would not yield significant size reduction.
Solution Approach 2:
The system implements feedback by measuring packet entropy and using this information to make informed compression decisions. The entropy measurement provides feedback about the compressibility of each packet, allowing the system to adapt its behavior and avoid applying compression to high-entropy data where it would consume CPU resources without achieving meaningful size reduction.
Data Source
AI summary
An approach for an adaptive, performance-oriented, and compression-assisted encryption scheme implemented on a host computer to adaptively improve utilization of CPU resources is provided. The method comprises queueing a new data packet and determining a size of the new data packet. Based on historical data, a plurality of already encrypted data packets is determined. Based on information stored for the plurality of already encrypted data packets, an average ratio of compression for the plurality of already encrypted data packets is determined. Based on the average ratio of compression, a throughput of compression value and a throughput of encryption value, a prediction whether compressing the new data packet will reduce a CPU load is derived. If it is determined that compressing the new data packet will improve utilization of the CPU resources, then a compressed new data packet is generated by compressing the new data packet.