Cluster Statistics Aggregation via Pre-established Authenticated Channels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In clustered network appliances, aggregating statistics across nodes is challenging due to active or inactive node changes, impacting performance, as existing methods struggle to efficiently collect and manage data between nodes.
Innovation Solution
A system where a cluster coordinator determines active nodes, establishes communication channels, and aggregates statistics by relaying requests and responses through established authenticated channels, maintaining open channels for efficient data collection and minimizing the need for new channel establishment during node changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If communication channels are established for each statistics request, then authentication and security are ensured, but the overhead of channel establishment impacts performance
Solution Approach 1:
The system establishes communication channels between cluster nodes in advance, before statistics aggregation requests are made. These pre-established channels remain open and can be reused for multiple statistics requests, eliminating the need to re-establish channels for each request while maintaining authentication security.
2Adaptability or versatility
If new communication channels are established when nodes become active or inactive, then the cluster topology is updated, but the channel establishment process impacts performance
Solution Approach 1:
The system dynamically maintains communication channels by detecting when cluster nodes become active or inactive. When topology changes occur, the system updates the set of active nodes and adjusts the communication channel configuration accordingly, ensuring channels are maintained only with currently active nodes without requiring complete channel re-establishment.
3Loss of information
If statistics are collected from all cluster nodes, then comprehensive cluster statistics are obtained, but the data collection process impacts performance
Solution Approach 1:
The system introduces an intermediary component that coordinates statistics collection across cluster nodes. This intermediary manages the aggregation process by collecting statistics from individual nodes through pre-established communication channels and consolidating the data, thereby reducing the performance impact on individual nodes while ensuring comprehensive statistics are gathered.
Data Source
AI summary
For multiple multi-core nodes in a cluster, the filtered statistics clients contacts the aggregator on a master node of the cluster, referred to as the cluster configuration owner (“CCO”) or cluster coordinator and expects the stats aggregated from all the cluster nodes. The aggregator on the CCO nodes relay the client request to packet engines on the CCO node and to an aggregator on each of the other nodes in the cluster. Then the CCO node aggregator gets responses from other cores on the node and responses from all other cluster node aggregators. The CCO node aggregator aggregates the responses and sends back the aggregated response to the clients. Communication between nodes is via a static authenticated communication channel.