Switch Offloading for Data Replication Latency

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Complex computing environments in network operators and communication service providers face performance issues due to latencies in write operations for large data storage and increased processing demands in service function chains, particularly in cloud and high-performance computing environments.

Innovation Solution

A system that offloads data object replication and service function chain management to a switch, utilizing hardware offload capabilities for replication and metadata processing, reducing latency and overhead by performing these tasks within the switch rather than individual computing nodes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If data object replication is performed across multiple storage nodes to provide redundant storage and data recovery, then storage reliability is improved, but write operation latency increases

Engineering Contradiction:

Improvestorage reliabilityVSAvoidwrite operation latency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent extracts the data replication function from the storage nodes and places it in the network switch. The switch independently handles replication of data objects to multiple storage nodes without requiring coordination or processing at the storage nodes themselves, thereby reducing write operation latency while maintaining storage reliability through redundant copies.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The network switch acts as an intermediary between the write operation source and the storage nodes. It receives write operations, performs replication to multiple storage nodes, and manages the coordination of redundant storage without involving the storage nodes in the replication decision-making process, thus improving write performance while ensuring data reliability.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If service function chain management is performed inside containers or virtual machines in virtualized environments, then scalability is improved, but processing performance deteriorates as load increases

Engineering Contradiction:

ImprovescalabilityVSAvoidprocessing performance

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent replaces the software-based service function chain management (running in containers or VMs) with hardware-based implementation in the network switch. This substitution of mechanical/software system with hardware system eliminates the performance degradation associated with increased processing loads in virtualized environments while maintaining scalability.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Speed

If hardware offload capabilities are used for replication and metadata processing in the switch, then processing speed is improved, but device complexity increases

Engineering Contradiction:

Improveprocessing speedVSAvoidswitch complexity

Core Design Contradiction:

SpeedVSDevice complexity

Solution Approach 1:

The network switch is designed with multi-functionality, combining data replication, metadata processing, and service function chain management capabilities within a single device. This universal approach allows the switch to handle multiple tasks that would otherwise require separate processing systems, improving overall processing speed while managing complexity through integrated design.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS11082515B2Technologies for offloading data object replication and service function chain management

Publication Date: 2021.08.03 INTEL CORP

AI summary

Technologies for offloading data object replication and service function chain management include a switch communicatively coupled to one or more computing nodes capable of executing virtual machines and storing data objects. The switch is configured to determine metadata of a service function chain, transmit a network packet to a service function of the service function chain being executed by one or more of the computing nodes for processing the network packet. The switch is further configured to receive feedback from service function, update the metadata based on the feedback, and transmit the network packet to a next service function of the service function chain. Additionally or alternatively, the switch is configured to identify a plurality of computing nodes (i.e., storage nodes) at which to store a received data object, replicate the data object based on the number of storage nodes, and transmit each of the received data object and replicated data object(s) to different corresponding storage nodes. Other embodiments are described and claimed.