Host Timestamp Mirroring for Storage Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In data mirroring for disaster recovery, timestamp discrepancies between hosts and primary storage controllers can hinder rapid recovery, as mirrored copies timestamped with the primary storage controller's time may not accurately reflect the host's time, requiring log file examination and slowing down disaster recovery processes.
Innovation Solution
The method involves timestamping mirrored copies with both the host timestamp and the elapsed time from the last host I/O command, along with accounting for potential clock drift in the primary storage controller, ensuring that recovery operations are aligned with the host's time, thus facilitating faster disaster recovery by directly using the host timestamp.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If mirrored copies are timestamped with the primary storage controller's time, then the mirroring process is simple to implement, but disaster recovery is slowed down due to timestamp discrepancies requiring log file examination
Solution Approach 1:
The patent introduces an intermediary mechanism that captures and transfers host timestamps to the secondary storage controller. This mediator (the timestamp transfer mechanism) bridges the timekeeping systems of the host and storage controller, allowing mirrored copies to be accurately timestamped with host time without requiring complex log file examination during recovery.
Solution Approach 2:
The patent applies preliminary action by pre-transferring host timestamps to the secondary storage controller before disaster recovery is needed. The timestamps are captured and stored in advance with the mirrored copies, so when recovery is required, the system can immediately use these pre-prepared timestamps without needing to examine log files or perform time synchronization calculations.
2Loss of time
If host timestamps are captured and transferred to the secondary storage controller, then disaster recovery speed is improved, but the complexity of the mirroring system increases
Solution Approach 1:
The patent makes the storage mirroring system multi-functional by enabling it to both mirror data and capture/transfer timestamps using the same infrastructure. The secondary storage controller performs multiple functions: storing mirrored copies, receiving host timestamps, and using these timestamps for rapid recovery. This universal approach avoids adding separate dedicated timestamp management systems.
Solution Approach 2:
The system applies self-service by having the secondary storage controller automatically receive, store, and utilize host timestamps without requiring external timestamp management systems. The mirroring process itself serves the dual purpose of data replication and timestamp capture, eliminating the need for separate timestamp synchronization mechanisms.
3Device complexity
If only the primary storage controller's clock is used for timestamping, then the system maintains simple timekeeping, but accuracy relative to host time deteriorates due to clock drift
Solution Approach 1:
The patent uses preliminary action by capturing the host timestamp at the moment the I/O command is received, before any clock drift can occur between the host and storage controller. This pre-captured timestamp is then transferred to the secondary storage controller, ensuring accurate time recording without relying on the storage controller's potentially drifted clock.
Solution Approach 2:
The patent replaces the mechanical clock-based timestamping system of the storage controller with a software-based timestamp capture mechanism that records the host's timestamp directly. This substitution eliminates the need for physical clock synchronization and avoids the drift issues inherent in independent hardware clocks.
Data Source
AI summary
A primary storage controller receives an input/output (I/O) command from a host, wherein a host timestamp is associated with the I/O command. During a mirroring of storage volumes to a secondary storage controller, the primary storage controller communicates the host timestamp associated with the I/O command to the secondary storage controller, wherein mirrored copies of the storage volumes are timestamped based on at least the host timestamp and an elapsed time since a last host I/O command. A recovery is made from a failure of one or more of the storage volumes in the primary storage controller, by using the timestamped mirrored copies of the storage volumes.