Asynchronous Replication Using Loose Global Clock
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Prior asynchronous replication services face scalability and latency issues when handling multiple write interception components, especially in configurations spanning multiple geographical sites, due to the need for extremely accurate hardware-based global clocks which are expensive and impractical for distributed environments.
Innovation Solution
A write ordering style asynchronous replication service utilizing a loosely-accurate global clock, where each source replication node intercepts writes, generates metadata with a Time Identifier, and buffers them for propagation to target replication nodes, ensuring consistent replication across multiple geographical sites without the need for expensive hardware-based solutions.
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) write interception and buffering at the source with immediate acknowledgment, and (2) asynchronous propagation to target systems. This segmentation allows the source to acknowledge writes without waiting for target confirmation, reducing latency while maintaining data consistency through sequential application at the target.
Solution Approach 2:
The write interception component performs preliminary actions by intercepting and buffering writes at the source before propagation. The metadata including sequence numbers and timestamps is prepared in advance, enabling the target to apply writes in the correct order without blocking the source acknowledgment, thus resolving the contradiction between speed and consistency.
2Productivity
If multiple write interception components are deployed to improve scalability, then scalability is improved, but write ordering consistency becomes difficult to maintain
Solution Approach 1:
The patent introduces metadata as an intermediary mechanism that carries ordering information (sequence numbers, timestamps, error boundaries) from multiple write interception components to the target. This metadata acts as a mediator that enables the target to correctly order and apply writes from multiple sources without requiring complex coordination between the interception components themselves, thus maintaining scalability while ensuring consistency.
Solution Approach 2:
The patent changes the parameter representation by attaching rich metadata (sequence numbers, timestamps, error boundaries) to each write operation. This parameter enrichment allows the target to determine write order based on these parameters rather than relying on synchronized clocks or complex inter-component communication, enabling multiple interception components to operate independently while maintaining global ordering consistency.
3Measurement precision
If hardware-based global clocks are used to maintain write ordering, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces expensive hardware-based global clocks with software-generated metadata that includes timestamps and sequence numbers. These metadata objects are inexpensive to create and manage, and their disposable nature (being regenerated with each write) allows for simple software-based time tracking without requiring complex hardware synchronization infrastructure, thus reducing device complexity and cost while maintaining sufficient precision for ordering.
Solution Approach 2:
The patent substitutes the mechanical/hardware-based global clock system with a software-based metadata approach. Instead of relying on physical clock synchronization hardware across distributed systems, the solution uses software-generated sequence numbers and timestamps embedded in write metadata, replacing complex hardware mechanisms with simpler software logic that achieves the same ordering function.
4Productivity
If writes are propagated immediately to target storage space, then productivity is improved, but data loss risk increases if source fails before propagation
Solution Approach 1:
The patent implements beforehand cushioning by having the write interception component buffer intercepted writes locally at the source before propagation to the target. This buffer acts as a cushion that protects against data loss: if the source fails before propagation, the buffered writes can be recovered and re-propagated. This allows immediate acknowledgment to the source (maintaining productivity) while ensuring data safety through the buffer cushion.
Solution Approach 2:
The write interception component performs self-service by autonomously buffering writes and managing their propagation without requiring continuous source-target coordination. The component can independently recover from failures by re-propagating buffered writes, and the target can independently apply writes in order based on metadata, enabling both systems to operate autonomously while maintaining data safety and productivity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A write ordering style asynchronous replication service utilizes a loosely-accurate global clock in a data backup and recovery storage environment. The storage environment includes a set of source storage systems illustratively embodied as source replication nodes 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 configured to service the consistency group. The write ordering style service utilizes the loosely-accurate global clock to provide consistent replication of a storage space of the consistency group.