Quick verdict
- Xpenology on Proxmox is great for learning DSM and experimenting without buying real hardware — but not suitable for important data
- SHR is the best default for homelab if drives are different sizes
- RAID 5 is fine when all drives are the same size and you want to stay RAID-standard
- SHR-2 or RAID 6 when drives are old or the data genuinely matters
The most important takeaway from this article: when using SHR, always use NAS drives — the reason is explained in detail below.
Real Synology hardware vs Xpenology on Proxmox
These are two fundamentally different ways to run DSM.
Real Synology (official hardware)
DSM running on genuine Synology NAS: DS923+, DS1522+, DS1821+...
Pros:
- Fully supported by Synology — DSM updates won't brick the device
- Hardware optimized for NAS: power-efficient chip, low noise, compact form factor
- SMART, wake-on-LAN, fan control all work properly
- Warranty and RMA if hardware fails
- Full access to Surveillance Station, Hybrid Share, Active Backup for Business
Cons:
- Expensive: DS923+ runs around $600–700 USD before drives
- Limited hardware flexibility — can't freely upgrade CPU/RAM (except on some models)
- Locked into the Synology ecosystem
Xpenology on Proxmox (emulated DSM)
DSM running as a VM on Proxmox using the Arc bootloader, emulating a specific Synology model.
For the full installation walkthrough, see the Xpenology on Proxmox VE 8.2.4 guide.
Pros:
- No need to buy a NAS — runs on an existing PC or server
- Flexible resources: adjusting RAM or CPU is just a VM settings change
- Try DSM and learn its features before committing to real hardware
- Proxmox snapshots: roll back the entire VM if a DSM update goes wrong
Cons:
- Not supported by Synology — no warranty, some enterprise features unavailable
- Depends on the community-maintained Arc bootloader — if Synology changes the bootloader signature, DSM updates may break
- HDD passthrough through Proxmox requires a CPU with VT-d (IOMMU)
- If the Proxmox host has issues, the NAS VM goes down with it
- Not suitable for genuinely important data
When Xpenology on Proxmox makes sense:
- Lab environment for learning and experimenting with DSM
- Storing non-critical data: media, secondary backups
- Trying DSM before investing in real hardware
RAID comparison on Synology
Quick comparison table
| RAID | Min drives | Fault tolerance | Example: 3 × 4TB | Example: 1TB+2TB+3TB | Add mismatched drive |
|---|---|---|---|---|---|
| RAID 0 | 2 | 0 drives | 12TB | 6TB | Not supported |
| RAID 1 | 2 | n-1 drives | 4TB | 1TB | Not supported |
| RAID 5 | 3 | 1 drive | 8TB | 3TB ❌ wastes 3TB | Difficult, same size needed |
| RAID 6 | 4 | 2 drives | 4TB (needs 4 drives) | — | Difficult, same size needed |
| SHR | 1+ | 1 drive | 8TB | 4TB ✅ fully utilized | Add any size |
| SHR-2 | 4+ | 2 drives | 4TB (needs 4 drives) | — | Add any size |
RAID 0 — speed, no protection
Data is striped evenly across all drives. No parity, no redundancy.
| Config | Usable | Lose 1 drive |
|---|---|---|
| 2 × 4TB | 8TB | Total data loss |
| 3 × 4TB | 12TB | Total data loss |
| 1TB + 2TB + 3TB | 6TB | Total data loss |
Use when: scratch disk, temp storage, cache — data loss is acceptable. Never for anything important.
Cannot be expanded after creation.
RAID 1 — mirroring, safe, most storage-expensive
Every drive holds an identical copy. Usable = capacity of the smallest drive, regardless of how many drives there are.
| Config | Usable | Can lose |
|---|---|---|
| 2 × 4TB | 4TB | 1 drive |
| 3 × 4TB | 4TB | 2 drives |
| 2TB + 4TB | 2TB (capped at smallest) | 1 drive |
Use when: only 2 drives and maximum protection is the goal, or a system volume that needs simple redundancy.
RAID 5 — balance between capacity and protection
Data and parity are striped across all drives. One drive can fail and be rebuilt. Formula: (n-1) × smallest_drive_size.
| Config | Usable | Parity overhead | Notes |
|---|---|---|---|
| 3 × 4TB | 8TB | 4TB | Efficient |
| 4 × 4TB | 12TB | 4TB | Efficient |
| 1TB + 2TB + 3TB | 3TB | 3TB | Wastes 3TB — capped at smallest |
| 2TB + 2TB + 4TB | 6TB | 2TB | 4TB drive only uses 2TB |
The problem with mixed sizes: RAID 5 is capped by the smallest drive. With 1TB + 2TB + 3TB, the 2TB and 3TB drives can only contribute 1TB each to the parity-protected space — the remaining capacity is wasted. The result is 3TB usable from 6TB total.
Use when: all drives are the same size, or you need a standard RAID format that's readable on non-Synology hardware.
RAID 6 — survives two simultaneous drive failures
Like RAID 5 but with two parity stripes. Requires at least 4 drives.
| Config | Usable | Parity overhead | Can lose |
|---|---|---|---|
| 4 × 4TB | 8TB | 8TB | 2 drives |
| 5 × 4TB | 12TB | 8TB | 2 drives |
| 6 × 4TB | 16TB | 8TB | 2 drives |
Use when: drives are old, or the array is large. A RAID 5 rebuild on a 20TB+ array can take 1–2 days — if another drive fails during that window, all data is lost. RAID 6 buys an extra layer of protection during rebuild.
SHR — Synology's default recommendation
SHR partitions each drive into equal-sized segments and combines them using RAID-5 style protection (or RAID-1 for 2-drive setups), extracting usable capacity from leftover space that RAID 5 would discard.
How SHR calculates capacity — concrete examples
Case 1: All drives the same size (4TB + 4TB + 4TB)
Segment A: 4TB from drive 1 + 4TB from drive 2 + 4TB from drive 3
→ RAID-5: 8TB usable, 4TB parity (= size of largest drive)
With equal-size drives, parity overhead = one drive's worth of capacity. Result is identical to RAID 5.
Case 2: Mixed sizes (1TB + 2TB + 3TB)
SHR segments the drives in steps, reusing whatever capacity is left:
Step 1 — take 1TB from each drive:
Segment A: 1TB (d1) + 1TB (d2) + 1TB (d3) → RAID-5 = 2TB usable
Remaining: d1=0, d2=1TB, d3=2TB
Step 2 — take 1TB from the two remaining drives:
Segment B: 1TB (d2) + 1TB (d3) → RAID-1 = 1TB usable
Remaining: d3=1TB (alone — no pair available, no parity protection)
Segment C: 1TB (d3 solo) → 1TB usable, unprotected
Total: 2 + 1 + 1 = 4TB usable
Parity used: 2TB (not 3TB = the largest drive)
Warning: the last 1TB on the largest drive has no parity with only 3 drives
SHR vs RAID 5 with mixed drives:
| Config | Total | RAID 5 | SHR | Saved vs RAID 5 |
|---|---|---|---|---|
| 1TB + 2TB + 3TB | 6TB | 3TB | 4TB | +1TB |
| 2TB + 2TB + 4TB | 8TB | 6TB | 6TB | Same |
| 1TB + 3TB + 4TB | 8TB | 3TB | 5TB | +2TB |
| 2TB + 4TB + 6TB | 12TB | 6TB | 9TB | +3TB |
The bigger the size mismatch between drives, the more SHR recovers over RAID 5.
Rules when adding a drive to SHR
When adding a new drive to expand the pool, the new drive must be at least as large as the current largest drive to actually increase usable storage. For example:
- Current pool: 2TB + 4TB (SHR = 4TB usable)
- Add 2TB drive: pool becomes 2TB+4TB+2TB — usable barely changes
- Add 4TB drive: pool becomes 2TB+4TB+4TB → usable grows to 6TB ✅
- Add 8TB drive: usable grows further ✅
Important limitation: SHR is Synology-proprietary. If the NAS hardware fails completely, you need another Synology NAS to read the data — the drives cannot be plugged into a standard PC.
SHR-2 — SHR that survives two failures
Same as SHR but keeps two parity segments instead of one, the same relationship as RAID 5 vs RAID 6. Parity overhead = 2 × largest drive.
| Config | Usable | Parity | Can lose |
|---|---|---|---|
| 4 × 4TB | 8TB | 8TB | 2 drives |
| 4 × 4TB + 1 × 2TB | 8TB+ | 8TB | 2 drives |
Minimum requirement: 4 drives
Use when: data is critical, drives are mixed sizes, or the array is large enough that a rebuild failure would be catastrophic.
Which RAID for homelab
Only 2 drives: → RAID 1. No other choice if data protection matters.
3–4 drives, same size: → RAID 5 if you want standard RAID. SHR if you're staying in the Synology ecosystem long-term.
3–4 drives, mixed sizes: → SHR is almost always the better choice.
4+ drives, critical data: → SHR-2 or RAID 6.
Only need speed, data is disposable: → RAID 0, but always keep a separate backup.
SHR requires NAS drives — not a suggestion
This is the thing most people skip when first setting up Synology, especially on Xpenology.
Why SHR expansion takes so long
When you add a new drive to an SHR pool, Synology must reshape the entire array — redistributing all data and parity across all drives, including the new one. This is not a simple operation. For a large array:
- Adding a 4TB drive: 12–24 hours
- Adding an 8TB drive: 24–48 hours
- Adding a 12TB+ drive: potentially 3–5 days
Monitor progress over SSH:
cat /proc/mdstat
# finish=847.8min speed=57623K/sec
For a detailed walkthrough of monitoring and precautions during expansion, see Adding a drive to Synology SHR.
Why desktop drives fail during long expansions
For the entire duration of the reshape, every drive in the array is being read and written continuously without rest. Desktop drives (WD Blue, Seagate Barracuda...) are designed for short workloads with a duty cycle of around 8 hours/day. Running them at full load for 24–48 hours straight:
- Raises temperatures beyond their design spec
- Increases read/write error rates
- Makes them more likely to fail mid-reshape — the worst possible scenario: the array is already degraded, and one more failure means total data loss
NAS drives (WD Red, WD Red Pro, Seagate IronWolf, Seagate IronWolf Pro) are designed for 24/7 duty cycles, optimized for long sequential workloads, and have firmware tuned to handle vibration from multiple drives in the same chassis.
This isn't excessive caution — it's a real failure mode. Many people run desktop drives in Synology fine under normal conditions, but a reshape running for days is exactly when a desktop drive is most likely to fail.
Practical guide
| Scenario | Drive recommendation |
|---|---|
| Lab, experimenting, non-critical data | Desktop drive is OK temporarily |
| Xpenology on Proxmox, personal data | NAS drive recommended |
| SHR with regular expansion | NAS drive required |
| Real Synology hardware, important data | NAS drive required, consider Pro series |
WD Red Plus (CMR) is the most popular choice for homelab NAS. Avoid older WD Red SMR drives (5400 series) for SHR arrays — SMR has poor random write performance, which significantly extends reshape time.
Checklist before creating a storage pool
- Chosen the right RAID type based on needs (number of drives, data importance, mixed sizes or not)
- Using NAS drives if planning to use SHR and expand later
- Checked SMART on every drive before adding to the pool (
synodiskport -disk sda -health) - Have a separate backup for important data — SHR and RAID are not backups
- If running Xpenology: HDD passthrough via IOMMU rather than virtual disks