Virtualized Grid Consistency Group for Data Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data protection systems face challenges in maintaining continuous data protection without slowing down the production site, as they struggle to keep pace with high data transaction rates, leading to potential shutdowns and significant data loss in the event of a disaster.
Innovation Solution
The implementation of a virtualized grid consistency group that replicates a logical unit across multiple data protection appliances, allowing for efficient distribution of IOs and journaling across different offsets, enabling dynamic adaptation to changing data transaction rates and minimizing downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If journaling is used to enable continuous data protection and rollback capability, then the ability to recover to any specified point in time is improved, but the overhead of multiple data transactions at the backup site causes the backup site to fall behind high data transaction rates at the production site
Solution Approach 1:
The logical unit is divided into multiple strips that are distributed across different backup storage systems. Each strip is journaled independently, allowing parallel processing of journaling operations. This segmentation enables the backup site to handle high data transaction rates by distributing the journaling workload across multiple systems rather than a single bottleneck.
Solution Approach 2:
The patent introduces a new dimension of parallelism by distributing journaling operations across multiple backup storage systems simultaneously. Instead of sequential journaling on a single system, the invention creates a multi-dimensional journaling architecture where multiple DPAs process journal entries in parallel, significantly increasing the throughput capacity for maintaining continuous data protection.
2Reliability
If data replication is used to create copies of production site data on backup storage systems, then data protection capability is improved, but the backup site may not be able to keep pace with high data transaction rates, causing a backlog of un-logged data transactions
Solution Approach 1:
The logical unit is segmented into multiple strips distributed across different backup storage systems. This segmentation allows the backup site to process data transactions in parallel across multiple systems, eliminating the single-point bottleneck that causes delays. Each DPA handles a portion of the journaling workload independently, maintaining pace with high data transaction rates.
Solution Approach 2:
The system performs preliminary striping and distribution of data before journaling operations begin. By pre-organizing data into distributed strips across multiple DPAs, the system is prepared to handle incoming data transactions efficiently without causing backlogs. This preliminary organization enables immediate parallel processing capability.
3Reliability
If continuous data protection is implemented with journaling, then the ability to rollback to previous points in time is improved, but the backup site is forced to slow down production site transactions when it cannot finish backing up one data transaction before the next occurs
Solution Approach 1:
The journaling workload is segmented and distributed across multiple backup storage systems. Each DPA independently journals its assigned strips, allowing parallel processing that maintains high data transaction speeds. This eliminates the need to slow down production transactions, as the distributed architecture provides sufficient throughput capacity to handle the journaling overhead without impacting production performance.
Data Source
AI summary
In one aspect, a method includes forming a virtualized grid consistency group to replicate a logical unit, running a first grid copy on a first data protection appliance (DPA) replicating a first portion of the logical unit, running a second grid copy on a second DPA replicating a second portion of the logical unit, sending IOs to the first DPA if the IOs are to a first set of offsets and sending IOs to the second DPA if the IOs are to a second set of offsets.