Connection Bucketing for Asymmetric Clustered Multiprocessor Failover

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Mirroring between computing devices in clustered systems is challenging, especially when individual devices within a chassis fail, as existing solutions do not effectively redistribute connections to ensure high availability and continuity of service.

Innovation Solution

Implementing a disaggregator (DAG) that routes incoming connections based on hash algorithms and attributes, allowing buckets to be dynamically reassigned to healthy computing devices, ensuring consistent routing even if devices fail, and using a similar hashing algorithm across primary and failover chassis to maintain connection integrity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If computing devices are mirrored between primary and failover chassis to improve availability, then system reliability is improved, but connection redistribution complexity increases when individual devices fail

Engineering Contradiction:
Improvesystem availabilityVSAvoidconnection redistribution complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments connections into buckets based on hash algorithms, allowing individual computing devices to fail without affecting all connections. Each bucket can be independently reassigned to healthy devices, reducing the complexity of connection redistribution during failures.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a connection bucketing mechanism as an intermediary between connections and computing devices. This bucketing layer absorbs the complexity of failure redistribution, shielding the system from direct device failure impacts and enabling simpler failover operations.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If connection bucketing with hash algorithms is implemented to enable dynamic reassignment, then connection integrity is maintained, but routing decision complexity increases

Engineering Contradiction:
Improveconnection integrityVSAvoidrouting decision complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent performs preliminary bucket assignment using hash algorithms before failures occur. This pre-computed bucket mapping enables fast failover decisions without complex real-time routing calculations, maintaining connection integrity while simplifying runtime routing decisions.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent changes the routing parameter from direct device addressing to bucket-based indirect routing. By transforming connection identifiers into bucket assignments through hashing, the system achieves deterministic routing while reducing the complexity of dynamic reassignment operations.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If dynamic reassignment of buckets to healthy devices is implemented, then high availability is achieved, but system state management complexity increases

Engineering Contradiction:
Improvehigh availabilityVSAvoidsystem state management complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements self-service bucket reassignment where healthy computing devices automatically claim unassigned buckets when failures occur. This eliminates the need for complex centralized coordination and manual state management, enabling autonomous failover while maintaining high availability.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent introduces dynamic bucket-to-device mapping that adapts to failure conditions. The system transitions from static one-to-one mappings to dynamic many-to-many relationships, enabling flexible resource allocation while simplifying state management through automated redistribution protocols.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS9497285B1Connection bucketing in mirroring asymmetric clustered multiprocessor systems
Publication Date: 2016.11.15 F5 NETWORKS INC
  • US9497285B1 patent drawing
  • US9497285B1 patent drawing
  • US9497285B1 patent drawing

AI summary

Creating a connection between one of a first plurality of computing devices in a primary chassis and one of a second plurality of computing devices in a failover chassis. A first plurality of buckets may be associated with the primary chassis, a second plurality of buckets may be associated with the failover chassis, where the first plurality of buckets may correspond to the second plurality of buckets. One of the first plurality of computing devices may be associated with one of the first plurality of buckets, and can create a connection with attributes such that a disaggregator in the failover chassis routes the connection to one of the second plurality of computing devices, wherein the one of the second plurality of computing devices may be associated with a bucket of the second plurality of buckets that corresponds to the one of the first plurality of buckets.