Distributed Virtual Switch Service Node Placement

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Data center networks face challenges in optimizing resource utilization and reducing latency in distributed service chains due to dynamic node placement and non-deterministic service chain latencies, which affect application performance and availability.

Innovation Solution

A method that involves a virtual supervisor module receiving inter-node latency information from packet headers, modifying service node locations within a distributed virtual switch to reduce latency, and using virtual Ethernet modules to store and calculate runtime traffic latencies, thereby optimizing path history and service node placement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If service nodes are distributed across multiple locations in the data center network, then service chain flexibility and resource utilization are improved, but inter-node latency increases and performance becomes non-deterministic

Engineering Contradiction:

Improveservice chain flexibilityVSAvoidinter-node latency

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by measuring inter-node latency between service nodes before traffic actually flows through the service chain. These pre-measured latency values are stored and used to predict end-to-end service chain latency, enabling the load balancer to make informed placement decisions that balance flexibility with latency optimization.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback mechanisms where actual inter-node latency measurements are continuously collected from packet headers as traffic traverses service nodes. This feedback is used to update latency databases and refine future service chain placement decisions, creating a closed-loop system that adapts to changing network conditions while maintaining deterministic performance predictions.

Inventive Principle:

Principle #23Feedback

2Productivity

If service nodes are co-located to reduce inter-node latency, then throughput and deterministic performance are improved, but service chain distribution and resource utilization decrease

Engineering Contradiction:

ImprovethroughputVSAvoidservice chain distribution

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system applies dynamics by enabling service nodes to be dynamically placed and relocated based on measured latency characteristics and traffic patterns. Rather than fixed co-location, the load balancer can service chain service nodes to different physical or virtual locations depending on current network conditions, traffic demands, and latency measurements, optimizing throughput while maintaining manageable distribution complexity.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes parameters by using measured inter-node latency values as key parameters for service chain placement decisions. The load balancer selects service nodes and their locations based on latency parameters derived from actual traffic measurements, allowing flexible adjustment of service chain configurations to optimize throughput without requiring complex manual distribution planning.

Inventive Principle:

Principle #35Parameter changes

3Ease of operation

If traditional load balancing algorithms are used, then basic traffic distribution is achieved, but deterministic end-to-end latency prediction and optimization are not possible

Engineering Contradiction:

Improvetraffic distributionVSAvoidend-to-end latency prediction

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The system substitutes traditional mechanical load balancing algorithms with a latency-aware decision-making mechanism. Instead of using simple round-robin or hash-based distribution, the load balancer uses measured inter-node latency data to predict end-to-end service chain latency and make intelligent routing decisions, achieving deterministic latency prediction while maintaining ease of operation through automated measurements and calculations.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system introduces an intermediary mechanism in the form of a latency database and prediction algorithm that sits between the load balancer and service nodes. This intermediary collects latency measurements from packet headers, stores them in a database, and provides predictive latency estimates to the load balancer, enabling precise end-to-end latency prediction without complicating the overall system operation.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS9491094B2Path optimization in distributed service chains in a network environment

Publication Date: 2016.11.08 CISCO TECHNOLOGY INC

AI summary

An example method for path optimization in distributed service chains in a network environment is provided and includes receiving information about inter-node latency of a distributed service chain in a network environment comprising a distributed virtual switch (DVS), where the inter-node latency is derived at least from packet headers of respective packets traversing a plurality of service nodes comprising the distributed service chain, and modifying locations of the service nodes in the DVS to reduce the inter-node latency. In specific embodiments, the method further includes storing and time-stamping a path history of each packet in a network service header portion of the respective packet header. A virtual Ethernet Module (VEM) of the DVS stores and time-stamps the path history and a last VEM in the distributed service chain calculates runtime traffic latencies from the path history and sends the calculated runtime traffic latencies to a virtual supervisor module.