NFVI Storage Decoupling for Edge Cloud Management

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In mobile core networks, deploying telecommunication cloud networks with thousands of remote locations for Multi-Access Edge Computing (MEC) and Virtualized Radio Access Network (vRAN) faces challenges due to limited power, rack space, and storage constraints, leading to inefficient use of compute nodes, poor NFV performance, and high operational expenses due to storage failures and manual software management in unmanned locations.

Innovation Solution

A system with a management entity and remote compute entities that decouples storage from compute and network functions, centralizes storage, and uses CPU core fencing to optimize resource utilization, enabling 100% node usage for hosting telco cloud workloads, and employs a centralized management infrastructure to manage thousands of edge PODs, reducing storage needs and operational costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If storage is decentralized at each remote edge location, then data access speed is improved, but storage space requirements and failure risk increase

Engineering Contradiction:

Improvedata access speedVSAvoidstorage space requirements

Core Design Contradiction:

SpeedVSQuantity of substance

Solution Approach 1:

The system segments storage functionality into two parts: hot data storage remains decentralized at edge locations for fast access, while cold data storage is centralized at remote locations. This segmentation allows the system to maintain fast data access for frequently accessed data while reducing total storage requirements by archiving less frequently accessed data centrally.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces a new dimension to storage architecture by adding a centralized storage layer accessible via wide area network, transforming the traditional single-layer decentralized storage model into a multi-layer hybrid model that combines both decentralized and centralized storage dimensions.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

2Productivity

If compute nodes are dedicated solely to hosting workloads, then workload performance is improved, but management and monitoring capabilities deteriorate

Engineering Contradiction:

Improveworkload performanceVSAvoidmanagement and monitoring capabilities

Core Design Contradiction:

ProductivityVSEase of operation

Solution Approach 1:

The management entity is designed with multi-functionality, serving as both a centralized management platform for provisioning and monitoring edge compute entities, and as a virtualization monitoring service that collects metrics from edge locations. This universal design allows one system to perform multiple functions without requiring separate dedicated infrastructure.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent introduces virtualization monitoring agents as intermediary components deployed on edge compute entities. These agents mediate between the workload running on compute nodes and the centralized management entity, allowing management and monitoring functions to be performed without directly interfering with workload performance.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If CPU cores are shared dynamically among virtual machines, then resource utilization is improved, but deterministic performance for low-latency workloads deteriorates

Engineering Contradiction:

Improveresource utilizationVSAvoiddeterministic performance

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system applies local quality by assigning different CPU allocation strategies to different workloads: low-latency workloads receive dedicated CPU cores with guaranteed performance characteristics, while non-time-critical workloads can utilize dynamically shared CPU resources. This localized differentiation allows the system to maintain high resource utilization overall while ensuring deterministic performance where required.

Inventive Principle:

Principle #3Local quality

4Extent of automation

If manual software management is used in unmanned locations, then software update control is improved, but operational expenses and time consumption increase

Engineering Contradiction:

Improvesoftware update controlVSAvoidtime consumption

Core Design Contradiction:

Extent of automationVSLoss of time

Solution Approach 1:

The system implements self-service through automated software provisioning and update mechanisms. The management entity automatically provisions software images to edge compute entities and applies updates without requiring manual intervention, enabling the system to manage software in unmanned locations autonomously and efficiently.

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS11714672B2Virtual infrastructure manager enhancements for remote edge cloud deployments

Publication Date: 2023.08.01 CISCO TECHNOLOGY INC

AI summary

A system is provided that includes one management cluster to manage network function virtualization infrastructure (NFVI) resources lifecycle in more than one edge POD locations, where resources include hardware and/or software, and where software resources lifecycle includes software development, upgrades, downgrades, logging, monitoring etc. Methods are provided for decoupling storage from compute and network functions in each virtual machine (VM)-based NFVI deployment location and moving it to a centralized location. Centralized storage could simultaneously interact with more than one edge PODs, and the security is built-in with periodic key rotation. Methods are provided for increasing NFVI system viability by dedicating (fencing) CPU core pairs for specific controller operations and workload operations, and sharing the CPU cores for specific tasks.