Asynchronous Replication Using Loosely-Accurate Global Clock
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Group stamping asynchronous replication services face limitations in achieving zero Recovery Point Objective (RPO) due to interval delays and scalability issues, particularly when using a single write interception component across multiple geographical sites, leading to potential data loss and performance penalties.
Innovation Solution
Implementing a loosely-accurate global clock to synchronize storage requests among source replication nodes, allowing for consistent replication across multiple geographical sites without the need for tight coordination, thus reducing the quiescent penalty and enhancing scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If synchronous replication is used to ensure data currency, then data consistency is improved, but latency and propagation delay increase
Solution Approach 1:
The patent segments the replication process into two independent phases: (1) source storage system processes writes and assigns timestamps, (2) target storage system receives and applies writes based on timestamp ordering. This segmentation allows the source to operate asynchronously without waiting for target confirmation, reducing latency while maintaining consistency through timestamp-based ordering.
Solution Approach 2:
The source storage system performs preliminary actions by pre-processing writes, assigning timestamps, and preparing replication data before the target system is ready to receive it. Writes are staged and buffered with timestamp metadata, allowing the target to apply them in correct order without blocking the source's write acknowledgment to applications.
2Speed
If asynchronous replication is used to reduce latency, then speed is improved, but data loss risk increases
Solution Approach 1:
The target storage system provides feedback to the source by reporting which timestamps have been successfully applied. The source uses this feedback to determine when to purge written data from buffers, ensuring that data is retained until confirmed applied at the target, thus preventing data loss while maintaining asynchronous operation.
Solution Approach 2:
The system cushions against data loss by maintaining buffered writes with timestamps at the source until the target confirms application. This creates a safety margin where data can be recovered if needed, protecting against the asynchronous nature's data loss risk while preserving speed benefits.
3Device complexity
If a single write interception component is used across multiple geographical sites, then device complexity is reduced, but scalability and performance deteriorate
Solution Approach 1:
Each source storage system independently performs write interception, timestamp assignment, and replication initiation without requiring a central coordinator. This self-service capability allows multiple geographically distributed source systems to operate autonomously, improving scalability while actually reducing overall system complexity by eliminating single points of failure and coordination overhead.
4Manufacturing precision
If group stamping is used to establish replication intervals, then manufacturing precision is improved, but loss of time increases due to interval delays
Solution Approach 1:
The patent changes the parameter for establishing replication groups from fixed time intervals to timestamp-based ordering. Each write receives a timestamp from the source's clock, and the target applies writes in timestamp order. This parameter change eliminates interval delays while maintaining precise ordering, as writes are grouped and applied based on their logical time stamps rather than rigid temporal intervals.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A group stamping style asynchronous replxcation servxce (100) utilizes a loosely- accurate global, clock (320) in a data backup and recovery storage environment. The storage environment includes a set of source storage systems illustratively embodied as source replication nodes (300) that cooperate to maintain a consistency group that may span multiple geographical sites. The storage environment also includes one or more target storage systems illustratively embodied as target replication nodes (400) configured to service the consistency group. The group stamping style service utilizes the loosely-accurate global clock (320) to provide consistent replication of a storage space, e.g., a target storage space, of the consistency group.