Replication Address Space Slices for Resource Throttling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The amount of data to be transmitted during each replication cycle in content-based storage systems is unknown, leading to inefficient use of system resources like network bandwidth and processing cycles, which can impact system performance.

Innovation Solution

Implementing a method that determines slices of a logical address space assigned to replication processors, calculates the expected number of slices to be replicated based on elapsed time and recovery point objective (RPO), and throttles replication by replicating additional slices if the actual number is less than expected, thereby optimizing resource usage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If data replication is performed continuously at high speed to meet RPO requirements, then data loss prevention is improved, but system resources (network bandwidth and processing cycles) are overconsumed

Engineering Contradiction:

Improvedata loss preventionVSAvoidsystem resource consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent implements dynamic throttling of replication operations based on real-time progress monitoring. The system adjusts replication speed dynamically by comparing actual replication progress against expected progress calculated from RPO requirements and elapsed time, thereby optimizing resource usage while ensuring RPO compliance

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system employs feedback mechanisms by continuously monitoring replication progress and using this information to adjust subsequent replication actions. The expected number of slices to be replicated is calculated based on feedback from elapsed time and RPO parameters, creating a closed-loop control system that prevents both over-replication and under-replication

Inventive Principle:

Principle #23Feedback

2Reliability

If replication speed is increased to ensure data is transmitted within RPO window, then reliability is improved, but system performance degrades due to resource contention

Engineering Contradiction:

ImproveRPO complianceVSAvoidsystem performance

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent applies partial action by replicating only the necessary number of slices calculated based on RPO requirements and actual progress, rather than continuously replicating at maximum speed. This prevents excessive resource consumption while ensuring sufficient replication to meet RPO targets

Inventive Principle:

Principle #16Partial or excessive action

3Productivity

If replication operations are throttled to reduce resource consumption, then system performance is improved, but risk of missing RPO deadline increases

Engineering Contradiction:

Improvesystem performanceVSAvoidRPO compliance

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system performs preliminary calculations to determine the expected number of slices that should have been replicated by now, based on RPO requirements and elapsed time. This advance planning allows the system to throttle replication appropriately while knowing exactly when to accelerate to meet the RPO deadline

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10409520B1Replication of content-based storage using address space slices

Publication Date: 2019.09.10 EMC IP HLDG CO LLC

AI summary

A computer program product, system, and method for determining one or more slices of a logical address space assigned to replication processor; determining an elapsed time since a start of a replication cycle; determining an expected number of slices that should have been replicated based on the elapsed time; and replicating one or more slices of the logical address space in response to determining the expected number of slices that should have been replicated is less than an actual number of slices replicated by the replication processor within the replication cycle.