Element 13.19
Storage and data durability on reused disks
- Sheet
- 13.19
- Chapter
- 13 Digital infrastructure
- State
- Drafted
- Release
- 16 November 2026
- Editor
- Contributed by Praecise as open research
- Reviewers
- To be named at review
- Safety
- S1, as the chapter
- Licence
- CC BY-SA 4.0
- Resources
- 8
The element
- Requirement
The community's data survives the failure of any single disk or machine on the storage itself, damaged data is detected rather than silently served, and backups follow the 3-2-1 rule.
- How to tell
- The storage layout tolerates at least one disk failure. In ZFS, a mirror of N disks survives N-1 failures and RAID-Z with P parity disks survives P failures (zpoolconcepts: overview of ZFS storage pools).
- Data is checksummed: ZFS checksums all data and metadata and repairs bad data from a good copy where redundancy exists (zpoolconcepts: overview of ZFS storage pools).
- Disk health is monitored continuously (smartmontools).
- Backups meet the 3-2-1 rule as CISA states it: three copies of important files, on two different types of storage media, with one copy stored off site; with scheduled recovery tests and the ability to roll back at least seven days (Back Up Business Data).
- Options
- ZFS on one machine. Mirrors, RAID-Z with one to three parity disks, dRAID with distributed spares and faster rebuilding, and hot spares (zpoolconcepts: overview of ZFS storage pools). OpenZFS recommends at least 2 GB of memory and calls ECC memory highly recommended (OpenZFS FAQ).
- Ceph across several machines. Distributed storage across hosts; its guidance states that more, smaller nodes are safer than fewer, denser ones, and advises bonded links across two switches (Ceph hardware recommendations).
- Hardware RAID controllers. Proxmox states that neither ZFS nor Ceph is compatible with a hardware RAID controller (Proxmox VE System Requirements); use either a hardware RAID or one of these, not both.
- Reuse of disks. Memory and storage are a large share of embodied carbon in the systems studied (Sustainability in HPC: Vision and Opportunities); disks that pass health checks are worth keeping in service, inside a redundant layout.
- Settled, open, contested
Settled. Redundancy is not backup; the 3-2-1 rule and tested restores are agreed guidance (Back Up Business Data). Checksumming file systems detect silent damage (zpoolconcepts: overview of ZFS storage pools). Open. How long reused disks of a given age last in community service is not documented in the sources opened. Contested. ZFS on one host against Ceph across several: ZFS is simpler to run; Ceph survives the loss of a whole host but asks for more machines, more network and more skill (Ceph hardware recommendations).
The four stages
- Learn: redundancy against backup: redundancy keeps service running through a disk failure; only a separate backup survives deletion, attack or fire (Back Up Business Data).
- Design: choose mirror or RAID-Z by disk count and tolerance (zpoolconcepts: overview of ZFS storage pools); allow the memory Proxmox advises for ZFS or Ceph (Proxmox VE System Requirements).
- Build: build the pool, enable disk monitoring (smartmontools), and set up backups to the 3-2-1 rule (13.5).
- Deploy: replace failing disks promptly, test restores on a schedule (Back Up Business Data).
Resources at each stage
1 Learn
Understand
- Back Up Business Data Guide, S1
- OpenZFS FAQ Guide, S1
- Sustainability in HPC: Vision and Opportunities Paper, S0
2 Design
Size and choose
- Ceph hardware recommendations Guide, S1
- Back Up Business Data Guide, S1
- Dell EMC PowerEdge R740 Technical Specifications: environmental specifications Guide, S0
- OpenZFS FAQ Guide, S1
- zpoolconcepts: overview of ZFS storage pools Guide, S1
- Proxmox VE System Requirements Guide, S0
4 Deploy
Install, run and measure
What changes between 30 adults and 300 people
At 30 adults one machine with a two-disk mirror and the backups of 13.5 is enough. At 300 people storage is either a larger RAID-Z pool with spares or Ceph across three or more hosts (Ceph hardware recommendations), with its own monitoring and a rota for disk replacement.
Reference designs
Open for contributions Designs with drawings, specifications and a bill of materials. Add to it
Parts list
Every resource that supports this element. Those marked External carry the label “External resource. Not made or verified by Public Protocol”.
| Resource | Type | Origin | Maturity | Safety | Licence, as stated at its source |
|---|---|---|---|---|---|
| Ceph hardware recommendations | Guide | External | Established | S1 | Not stated on the page opened |
| Back Up Business Data | Guide | External | Established | S1 | Not stated on the page opened |
| Dell EMC PowerEdge R740 Technical Specifications: environmental specifications | Guide | External | Established | S0 | Not stated on the page opened |
| OpenZFS FAQ | Guide | External | Established | S1 | Not stated on the page opened |
| zpoolconcepts: overview of ZFS storage pools | Guide | External | Established | S1 | Not stated on the page opened |
| Proxmox VE System Requirements | Guide | External | Established | S0 | Not stated on the page opened |
| smartmontools | Tool | External | Established | S0 | GNU General Public License, version 2 |
| Sustainability in HPC: Vision and Opportunities | Paper | External | Concept | S0 | Not stated on the page opened |
Depends on, and completes
It is one of 25 elements that complete Digital infrastructure.
Environments
No source found on how storage choices change by setting. Heat and humidity limits for the machines that hold the disks are in the server room element (Dell EMC PowerEdge R740 Technical Specifications: environmental specifications).
Safety
State and evidence
Drafted
An editor has written the element's text from the resources listed under it, citing them. It is checked when it is reviewed.
Evidence from a real place or system is recorded here, with what did not work. What the states mean
History
- Drafted in the release of 16 November 2026.