Shared Buffers for Network Processing Elements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing number of virtual machines (VMs) and queues supported by I/O devices leads to a significant requirement for hundreds of gigabytes of memory for receive buffers, resulting in high costs and power consumption, making current techniques for separately allocated receive buffers increasingly problematic.
Innovation Solution
Implementing encryption and decryption of data in shared buffers using exchanged block cipher keys, allowing multiple processing elements to share memory buffers and reducing the need for extensive physical memory allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If separate receive buffers are allocated to each queue for processing elements, then line rate throughput is sustained for each queue, but memory requirements increase to hundreds of gigabytes
Solution Approach 1:
The patent merges separate receive buffers for multiple queues into shared buffer pools that can be dynamically allocated to different queues. Instead of dedicating fixed memory to each queue, the system creates common buffer resources that multiple processing elements and queues can access, thereby reducing total memory requirements while maintaining throughput capability.
Solution Approach 2:
The patent implements dynamic buffer allocation where buffer assignments are not fixed but can be reassigned based on current queue activity and throughput requirements. This dynamic approach allows the same physical memory to serve multiple queues at different times, reducing the peak memory requirement from hundreds of GB to a fraction of that amount.
2Adaptability or versatility
If memory allocation is increased to support thousands of queues, then queue capacity is expanded, but power consumption and operational costs increase significantly
Solution Approach 1:
The patent makes memory buffers universal by creating shared buffer pools that can serve multiple queues and processing elements. The same physical memory resources perform multiple functions for different queues at different times, eliminating the need for dedicated memory per queue. This multi-functionality reduces total memory requirements from hundreds of GB to manageable levels, thereby reducing power consumption and operational costs.
3Speed
If dedicated receive buffers are assigned to each processing element, then data access speed is optimized, but memory cost and device complexity increase
Solution Approach 1:
The patent segments the buffer management system into distinct components: shared buffer pools, allocation managers, and queue-specific buffer identifiers. This segmentation allows multiple queues to access shared memory resources while maintaining logical separation and fast access paths. The segmentation enables efficient buffer management without requiring dedicated physical buffers for each queue, thus reducing complexity while preserving speed.
Data Source
AI summary
Examples are disclosed for exchanging a key between an input/output device for network device and a first processing element operating on the network device. Data having a destination associated with the first processing element may be received by the input/output device. The exchanged key may be used to encrypt the received data. The encrypted data may then be sent to a buffer maintained at least in part in a memory for the network device. The memory may be arranged to enable sharing of the buffer with at least a second processing element operating on the network device. Examples are also disclosed for the processing element to receive an indication of the storing of the encrypted data in the buffer. The processing element may then obtain the encrypted data from the buffer and decrypt the data using the exchanged key.