vSAN ESA objects display RAID 1 and RAID 5 components when a RAID 5 policy is applied to the object.
Article ID: 434061
Updated On:
Products
Issue/Introduction
Symptoms:
- In a VMware vSAN Express Storage Architecture (ESA) environment, the vSphere Client's Physical disk placement view displays both RAID 1 and RAID 5 components for storage objects, despite a RAID 5 (Erasure Coding) storage policy being applied.
Environment
VMware vSAN 8.x (Express Storage Architecture).
Cause
In vSAN ESA, the Log-Structured File System (vSAN LFS) splits every storage object into a Performance Leg and a Capacity Leg. The Performance Leg is always configured as a RAID 1 mirror to serve as a highly responsive landing zone for incoming writes. The Capacity Leg utilizes the assigned storage policy (e.g., RAID 5) for persistent storage. Details are below.
- The Performance Leg (RAID 1)
- Regardless of the storage policy applied, vSAN ESA creates a highly responsive, temporary landing zone for all incoming writes and metadata. To ensure incoming I/O is committed as quickly as possible while remaining fully redundant, vSAN ESA always configures this performance leg as a RAID-1 mirror across two hosts. This eliminates the traditional "write penalty" that usually comes with parity-based RAID configurations.
- Regardless of the storage policy applied, vSAN ESA creates a highly responsive, temporary landing zone for all incoming writes and metadata. To ensure incoming I/O is committed as quickly as possible while remaining fully redundant, vSAN ESA always configures this performance leg as a RAID-1 mirror across two hosts. This eliminates the traditional "write penalty" that usually comes with parity-based RAID configurations.
- The Capacity Leg (RAID 5)
- This is where the assigned storage policy takes effect. Once data lands safely in the RAID-1 performance leg, vSAN LFS organizes it and asynchronously destages it in full, efficient stripes down to the capacity leg.
The Concatenation we see at the top level is simply vSAN stitching these two legs together so the virtual machine sees a single, contiguous Hard disk. The active RAID-1 components handle the immediate, hot writes, while the RAID-5 components store the destaged, persistent data. This architectural design is how vSAN ESA achieves the high performance of RAID-1 alongside the space-saving efficiency of RAID-5.
Resolution
No corrective action is required. This is the expected, normal behavior for vSAN ESA to achieve RAID 1 performance alongside RAID 5 space-saving efficiency.
Additional Information
Note: The component count under RAID 5 depends on the available fault domains. Clusters with six or more fault domains list five components, while clusters with five or fewer list three components.
Some useful links:
- RAID-5/6 with the Performance of RAID-1 using the vSAN Express Storage Architecture
- vSAN Proof of Concept: vSAN Performance Testing (Check "Storage Policy" on Page 7).
- Capacity Overheads for the ESA in vSAN 8 (This article breaks down how this new two-tier object structure impacts overall cluster capacity. It explains that while the RAID-1 performance leg does consume some space, it remains relatively small and serves only to pipeline data efficiently into the persistent capacity leg.)
Feedback
