Home > Storage > PowerMax and VMAX > Data Protection > Dell PowerMax: Reliability, Availability, and Serviceability > Application I/O serviced by remote array
In the event of a redundant RAID failure on the R1 array, applications local to the R1 site will continue to access data through the remote R2 array. There may be some overhead on response time depending on distance, but the data remains available to the applications even though it cannot be read locally.
This core functionality has been an integral piece of SRDF for many years, originally providing the ability for the entire contents of a thick volume to be accessed from a remote array and rebuild data after a drive replacement even before local RAID schemes were implemented. In PowerMax arrays, which employ virtual provisioning, remote access is not required for an entire TDEV and is instead done on a per-track basis.
The local RAID group will be rebuilt from the remote array after replacement or recovery of the affected drives. An invoked spare drive can also participate in the rebuild in place of a RAID member. If the RAID members had failed simultaneously, both can be recovered and rebuilt concurrently. If the second drive failed while the first drive was rebuilding after being replaced, the remote data is used to finish rebuilding the data to the first drive, and the second drive will be rebuilt locally. Applications can remain online during the rebuild.
This functionality is available with SRDF/Metro, SRDF/S, and SRDF/A (when not in a spillover state). Being able to access data remotely, and rebuild RAID data from a remote array, enable SRDF configurations to take advantage of the efficiency and performance benefits of RAID 5 without incurring the performance penalties of RAID 6 protection. The Mean Time Between Part Replacement (MTBPR) of modern flash drives, and most replacements being proactive, make a dual failure very unlikely.
IDC tested the continuous operations without data loss from a total local RAID group failure with SRDF/Synchronous during overall validation of Dell VMAX All Flash arrays. The following is from the previously published IDC Lab Validation Report Dell VMAX All Flash: Essential Capabilities for Large Enterprise Mixed Workload Consolidation:
IDC Opinion: The active volume worked flawlessly during these fault injection tests, both upon initial failure and after re-establishing the failed resource… I/O was not interrupted, no data was lost, and in this case (with a relatively light workload) there was no long-term impact to overall system performance… Throughout the entire fault injection test, the array(s) continued to meet its specified Diamond Service Level.
Exploration of I/O Impact on Dual Local Drive Failure in a RAID 5 Configuration: