Packet Processing Hash Transfer for HA Cluster Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In high availability (HA) clusters, packet hashing operations performed by standby nodes are rendered useless due to encryption during encapsulation, leading to inefficient random routing by active nodes, as they cannot utilize the calculated hash values.
Innovation Solution
The calculated hash value from a standby node is carried in a peer divert header and transmitted to an active node, allowing the active node to decrypt and utilize it for load balancing, preserving security while optimizing routing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If encryption is applied during encapsulation to protect packet integrity, then security is improved, but the active node cannot access the calculated hash value for load balancing
Solution Approach 1:
The packet structure is segmented into encrypted and unencrypted portions. The hash value is placed in an unencrypted header section that the active node can access independently, while the payload remains encrypted for security. This segmentation allows simultaneous achievement of security and hash accessibility.
Solution Approach 2:
An intermediary header structure is introduced between the encrypted payload and the active node's processing. This header contains the hash value in an accessible format, acting as a mediator that provides the hash information without compromising the encrypted nature of the actual packet data.
2Productivity
If the standby node performs hashing operation on encrypted packets, then routing decisions can be made, but the hash value is wasted because the active node cannot see it
Solution Approach 1:
The hash value is calculated in advance by the standby node and preserved in the packet header before the packet reaches the active node. This preliminary action ensures that when the active node receives the packet, the hash value is already available and can be immediately used for load balancing without requiring recalculation.
Solution Approach 2:
The hashing operation initiated by the standby node continues to be useful throughout the packet's journey. By placing the hash value in an accessible location in the packet structure, the computational effort invested by the standby node is not wasted but rather continues to serve its purpose at the active node for load balancing decisions.
3Ease of operation
If the active node performs random load balancing without hash values, then routing simplicity is maintained, but routing optimization is lost
Solution Approach 1:
The packet itself carries the hash value information it needs for optimal routing within its own structure. The active node can autonomously access this pre-calculated hash value from the packet header and use it for load balancing decisions, combining the simplicity of direct access with the optimization of hash-based routing without requiring complex external systems.
Data Source
AI summary
The present technology carries a calculated hash value from a standby node to an active node in an HA cluster when the packet is redirected to the active node. The hash value calculated at the standby node can be placed in the peer divert header during inner Layer 3 processing and carried to the active node in a local data bus or data line. The active node can then decode the hash value received from the standby node using the divert header. In doing so, security is preserved for the transmitted packet, but the active node can also load balance the packet based on the hash value calculated by the standby node.