Backup glossary: plain definitions from A to Z

By Lucas Flores, IT Systeme Flores UG. Last updated 2026-10-22

0-9

3-2-1 rule #

Keep three copies of your data (the original and two backups) on two different kinds of storage, with one copy in a different location. It is a planning heuristic commonly attributed to the photographer Peter Krogh, not a standard. Read more: The 3-2-1 backup rule

3-2-1-1-0 rule #

The 3-2-1 rule plus one copy that is immutable or offline, and zero errors after a verified recovery test. It is an extension that Veeam describes and popularised, not a standard. Read more: The 3-2-1 backup rule

A

Air gap #

A copy that is not connected to the systems that could change or delete it, so a faulty job or an attacker on the live network cannot reach it. NIST describes an air gap as an interface where two systems are not connected physically and any logical connection is not automated. Read more: The 3-2-1 backup rule, Immutable backups, WORM, air gaps and ransomware. Source: NIST glossary: air gap

Append-only #

A mode in which new data can be added but existing data cannot be changed or deleted by the writer. Servers and clients, and the runs of file shares, back up to Restow append-only: the agent or the runner can add backups but never delete or overwrite them, and retention runs only on the Restow server. Read more: Server and client backup

Archive #

A long-term store of records that you keep because you must or want to, as opposed to a backup, which exists to restore after a loss. The Restow archive is an append-only mail store with a SHA-256 hash chain, search and export. Read more: Backup vs. archive, Mail import, export and the archive, GoBD and email archiving, explained

Audit log #

A record of who did what and when. Restow records every read and restore in a hash-chained, append-only audit log in every edition; the viewer and the export belong to Business and Service Provider. Read more: Microsoft 365 and IMAP backup

B

Backup chain #

The set of backup files a restore needs: a full backup and the incremental backups that depend on it. If one link is missing or damaged, every restore point after it can be affected. Read more: Full, incremental, differential and synthetic full backups compared

Backup job #

The configuration that says what is backed up, when, into which repository and for how long. Restow has jobs for mail and Microsoft 365 (Exchange Online, OneDrive, SharePoint sites and IMAP), for servers and clients, for file shares and for Proxmox VE guests (a preview). Read more: Microsoft 365 and IMAP backup

Backup window #

The period in which backups are allowed to run, usually chosen to limit the load on production systems and on the network. In Restow you can limit the bandwidth a backup may use within a time window that you set, and an agent policy can set a backup window in which scheduled backups of servers and clients may start. Read more: Microsoft 365 and IMAP backup

Bare-metal restore #

Restoring a whole machine, with its operating system, applications and settings, onto empty hardware from an image-level backup. Restow's agent backs up files only today; Linux image backup with bare-metal restore is planned. Read more: Server and client backup, Roadmap

Block-level backup #

A backup that reads a disk as blocks instead of as files, which suits virtual machine disks and whole-disk images. Restow's Proxmox VE backup works at block level for VM disks, in 4 MiB blocks; the agent for servers and clients is file-based. Read more: Proxmox VE backup (preview), Full, incremental, differential and synthetic full backups compared

Boot test #

A restore test that goes one step further than reading files back: the backup of a virtual machine or container is restored as a temporary copy, started in isolation and shown to boot. It proves that the restore point starts, not that the applications inside work. Restow's Proxmox VE boot test (a preview, off by default) restores a restore point as a new guest in a separate pool, isolates its network, starts it, waits for the guest agent and destroys it again. Read more: Proxmox VE backup (preview), Restore verification

C

Changed block tracking #

A mechanism in a hypervisor or storage layer that records which blocks of a disk changed since the last backup, so the backup reads only those. The name differs by vendor; QEMU, which Proxmox VE uses, calls its version a dirty bitmap. Read more: Full, incremental, differential and synthetic full backups compared

Checksum #

A short fingerprint computed from data. If even one byte changes, the checksum changes. Restow records a SHA-256 checksum of every item when it backs it up and compares it during restore checks. Read more: How Restow proves restorability

Chunk #

A piece of data that a backup tool stores and tracks on its own. Restow splits data into content-defined chunks of 256 KiB to 4 MiB (about 1 MiB on average; an object under 256 KiB is a single chunk), names each by its hash and packs them into pack files. Read more: Storage, Full, incremental, differential and synthetic full backups compared

Compliance mode (Object Lock) #

The stricter retention mode of S3 Object Lock: while the retention period runs, no user can overwrite or delete the locked object version, not even the root user of the account, and the period cannot be shortened. AWS names deleting the whole account as the only way around it before the date. Restow can use it for immutable backups, for 1 to 3650 days, and shows a restore point as immutable in green only when the storage itself reported a compliance retention. Read more: Immutable backups, WORM, air gaps and ransomware. Source: Amazon S3 User Guide: Locking objects with Object Lock

Content-defined chunking #

Splitting data at boundaries decided by the content itself instead of at fixed sizes, so inserting a few bytes moves only nearby boundaries and most chunks stay identical. FastCDC is a published algorithm for it, and Restow uses it. Read more: Full, incremental, differential and synthetic full backups compared. Source: FastCDC: a fast and efficient content-defined chunking approach for data deduplication (USENIX ATC 2016)

Copy repository #

Restow's term Restow's term for a repository with the role Copy: it receives every new backup in addition to the primary, older data is mirrored onto it and verified by hash during the weekly storage check. It covers mail, calendar, contacts, OneDrive, SharePoint sites, IMAP and Proxmox VE VM data; backups of servers, clients, file shares and Proxmox containers stay in the primary repository. A copy repository has its own immutability setting. Read more: Storage, The 3-2-1 backup rule

D

Data processor (GDPR Art. 28) #

A party that processes personal data on behalf of the controller and on its documented instructions, for example a hosting or storage provider that holds your backups. Article 28 requires a contract that covers, among other things, security measures, help with data subject requests and the deletion or return of the data at the end. Whether a given provider is a processor in your setup is for you and your data protection adviser to decide. Read more: GDPR and backups, Encryption and key handling. Source: Regulation (EU) 2016/679, Article 28 (EUR-Lex)

Deduplication #

Storing identical data once and referring to it from every place it occurs. How much it saves depends on the data. Restow deduplicates within a tenant and never across tenants. Read more: Full, incremental, differential and synthetic full backups compared, Storage

Delta query #

A Microsoft Graph request that returns only what was created, changed or deleted since the previous call, tracked with a state token. Restow uses it so that a Microsoft 365 backup run fetches only the changes. Read more: Microsoft 365 and IMAP backup. Source: Microsoft Learn: Use delta query to track changes in Microsoft Graph data

Differential backup #

A backup of everything that changed since the last full backup. Restoring needs the full backup plus the latest differential, and each differential grows until the next full backup. Read more: Full, incremental, differential and synthetic full backups compared. Source: Microsoft Learn: Differential backups (SQL Server)

Dirty bitmap #

A bit vector in which each flag marks a region of a disk that may have changed since tracking began, at a fixed granularity. Proxmox VE sets one up with a 4 MiB granularity for backup access, and Restow's Proxmox VE backup (a preview) reads only the changed blocks. Read more: Proxmox VE backup (preview), Full, incremental, differential and synthetic full backups compared. Source: QEMU documentation: Dirty bitmaps and incremental backup

E

Egress fee #

A charge by a cloud or storage provider for data that leaves its network, for example when you download a backup to restore it. Some providers include a traffic allowance, others bill per gigabyte. The EU Data Act ends switching charges, egress included, from 12 January 2027; read your contract for what ordinary downloads cost. Read more: Choosing backup storage. Source: European Commission: Data Act explained

Encryption at rest #

Encrypting data where it is stored, so that someone who can read the disk or the bucket cannot read the content. Restow encrypts every chunk with AES-256-GCM before it is written, with a separate key for each tenant, in every edition. Read more: Encryption and key handling

F

Failure domain #

A set of things that can fail together: one disk, one controller, one site, one provider account, one set of administrator credentials. Backup copies are only independent if they do not share a failure domain. Read more: The 3-2-1 backup rule

File share #

A folder on a file server that other computers reach over the network, usually over SMB or NFS. It holds working files that change all day, so a backup tool has to read it over the network, often without any software on the file server. Restow backs up SMB and NFS shares without an agent: it mounts the share read-only for one run and backs it up with restic. Read more: File shares (SMB and NFS), Full, incremental, differential and synthetic full backups compared

Forever-incremental #

A scheme with one full backup followed only by incremental backups, where older restore points are merged or pruned to meet retention, so no new full backup is ever taken. Veeam describes its forever forward incremental method this way. Read more: Full, incremental, differential and synthetic full backups compared. Source: Veeam: Backup methods

Forwarded copy (to the archive) #

A plain copy of a mail that the mail provider sends to another address while the mail is in its transport, set up with a BCC map (Postfix, mailcow), a Sieve redirect with :copy or a routing rule such as "Add more recipients" in Google Workspace. Restow's receiver accepts such copies as well as Exchange journal reports in Business and Service Provider, since 0.4.4. A forwarded copy has no journal envelope, so BCC recipients show only if the provider writes them into a header. Read more: Mail import, export and the archive

Full backup #

A backup of all selected data that is complete in itself. It is the simplest to restore and the largest and slowest to take. Read more: Full, incremental, differential and synthetic full backups compared

G

Garbage collection #

The cleanup that removes stored data which no restore point refers to any more, usually after retention has expired old restore points. In Restow it runs as part of the weekly storage check and re-packs data into new packs instead of changing a pack in place; on a repository with immutable backups, the space comes back only after the lock of the old data has run out. Read more: Storage

GFS (grandfather-father-son) #

A rotation that keeps daily, weekly and monthly, often also yearly, restore points and thins out older ones. It began as a scheme for rotating tape sets and is now usually a retention rule. Read more: Grandfather-father-son backup rotation

Governance mode (Object Lock) #

The milder retention mode of S3 Object Lock: most users cannot overwrite or delete a locked object version or change its lock, but a user with a special permission can bypass the lock by asking for it explicitly. It protects against mistakes and most accounts while leaving administrators a way out, and AWS suggests it for trying a retention setting before using compliance mode. Restow can use it for immutable backups too, never sends the bypass request, and shows such a lock in neutral, because any account with the bypass permission can lift it. Read more: Immutable backups, WORM, air gaps and ransomware. Source: Amazon S3 User Guide: Locking objects with Object Lock

Granular restore #

Restoring one item, such as a mail, a file or a folder, instead of a whole mailbox or machine. Restow offers granular and complete restores, non-destructive by default. Read more: Restore

H

Hash chain #

A sequence in which each entry's hash includes the hash of the entry before it, so changing an old entry breaks every later link. It makes tampering detectable; it does not prevent it. The Restow archive and audit log use a SHA-256 hash chain. Read more: Mail import, export and the archive, GoBD and email archiving, explained

I

Immutability #

The property that data cannot be changed or deleted for a set period, enforced by the storage rather than by good behaviour. In Restow it is a setting of each repository, off until an administrator turns it on: on S3 with Object Lock it locks backup data and archived message content for 1 to 3650 days, in compliance or governance mode, while a local disk or an NFS share gets only a deletion delay in Restow's own code, which is not immutability. It was tested against a simulated S3 service, not yet a real one. Read more: Immutable backups, WORM, air gaps and ransomware, The 3-2-1 backup rule, Immutable backups in Restow

Incremental backup #

A backup of what changed since the previous backup of any kind. It is small and fast, but a restore needs the last full backup and every incremental backup after it. Read more: Full, incremental, differential and synthetic full backups compared

J

Journaling (Exchange) #

An Exchange feature that sends a copy of messages matching a rule to a journaling mailbox outside Exchange Online. Microsoft recommends Microsoft 365 retention instead where that is possible. The Restow archive can receive journal reports over SMTP in Business and Service Provider. Journaling captures a mail in transport, before a user can change it; Restow labels that "At transport", against the "Near-time" of capture from the mailbox. Read more: Mail import, export and the archive, Backup vs. archive. Source: Microsoft Learn: Journaling in Exchange Online

K

Key encryption key (KEK) #

A key whose only job is to encrypt other keys. In Restow each tenant's data key is stored only in wrapped form, encrypted with the installation's master key, so the master key has to be kept safe separately from the server. Read more: Encryption and key handling

L

Legal hold #

An instruction that stops data from being deleted because of litigation or an investigation, overriding normal retention until it is lifted. Microsoft 365 offers Litigation Hold and eDiscovery holds; the Restow archive layer has legal hold per tenant or per mailbox in Business and Service Provider. Read more: Backup vs. archive, Does Microsoft 365 back up your data?, Email archiving under the GoBD

LTO tape #

Linear Tape-Open, a magnetic tape format from the LTO Program whose cartridges hold tens of terabytes and can be taken offline and stored away from the network. The LTO Program lists write-once (WORM) cartridges from LTO-3 and hardware encryption from LTO-4. Restow has no tape target. Read more: Choosing backup storage, Immutable backups, WORM, air gaps and ransomware. Source: LTO Program: benefits of LTO

M

Microsoft 365 Backup #

Microsoft's own paid backup product for SharePoint sites, OneDrive accounts and Exchange Online mailboxes, run from the Microsoft 365 admin center and billed per GB of protected data. It is separate from the recycle bins and retention that come with a Microsoft 365 subscription. Read more: Does Microsoft 365 back up your data?, Restow compared with Microsoft 365 Backup. Source: Microsoft Learn: Overview of Microsoft 365 Backup

MTD (maximum tolerable downtime) #

The total time the owner of a business process is willing to accept that process being disrupted, counting every kind of impact. The recovery time objective has to be shorter, because the time to reprocess data after a restore comes on top of it. NIST describes it in SP 800-34 Rev. 1. Read more: RPO and RTO explained. Source: NIST glossary: maximum tolerable downtime

N

NBD (Network Block Device) #

A protocol that lets a server export storage so that a client can use it as a block device. Proxmox VE exports VM disks over NBD to backup providers, and Restow's Proxmox VE helper reads the changed blocks that way. Read more: Proxmox VE backup (preview). Source: NBD protocol specification

Near-time capture #

Restow's term Capturing a mail for the archive from the mailbox shortly after delivery, for example through Microsoft Graph change notifications or an IMAP sync, instead of in transport. A mail that is moved or deleted before the capture fetches it can be missed, so near-time capture is not complete. "Near-time" is the label Restow's mail job editor gives these ways, next to "At transport" for journaling and forwarded copies; they are part of Business and Service Provider since 0.4.4. Read more: Mail import, export and the archive, Backup vs. archive

NFS (Network File System) #

A protocol that lets a machine use a directory on another machine, such as a NAS, as if it were local. Restow can use an NFS share as a repository, mounted by its mounter container, and can back up an NFS export as a file share without an agent. Read more: Choosing backup storage, Storage

Non-destructive restore #

A restore that writes to a new location or as new items and never overwrites the original. It is the default for every restore in Restow. Read more: Restore

O

Object Lock #

An Amazon S3 feature that keeps object versions from being overwritten or deleted for a fixed time or until a hold is removed, using a write-once-read-many model. Some S3-compatible services offer it as well. Restow checks a bucket's Object Lock setting when you turn on immutable backups for a repository, and refuses a bucket without it. Read more: Immutable backups, WORM, air gaps and ransomware, The 3-2-1 backup rule, Storage. Source: Amazon S3 User Guide: Locking objects with Object Lock

P

Pack file #

A file that bundles many chunks, up to 64 MiB in Restow, with an index at the end, so that storage holds fewer, larger objects instead of millions of small ones. Read more: Storage

Point-in-time restore #

Restoring data as it was at a chosen moment. It needs a restore point from that period, so how finely you can choose depends on how many restore points retention keeps. Read more: Grandfather-father-son backup rotation

R

Ransomware #

Malicious software that encrypts or deletes a victim's data and demands payment, usually for a key. A backup that the attacker can also reach or delete is no protection, which is why immutable storage and offline copies matter. Restow judges every new restore point for signs of ransomware from data the backup already has; it is a tripwire, not a virus scanner, a restore point without findings proves nothing, and it was tested with simulated attacks only. Read more: Immutable backups, WORM, air gaps and ransomware, Ransomware detection

Recoverable Items folder #

A hidden folder in every Exchange Online mailbox that holds items a user has deleted from the Deleted Items folder. By default they stay for 14 days, which an administrator can raise to 30; holds and retention policies change what is kept. Read more: Does Microsoft 365 back up your data?. Source: Microsoft Learn: Recoverable Items folder in Exchange Online

Recovery readiness #

Restow's term Restow's term for the proven state of a protected mailbox, OneDrive, SharePoint site, IMAP account, machine, file share or Proxmox VE guest. It is Ready only after a restore check has read the newest backup back and matched it, and a backup that has not been checked shows as Unverified, never green by default. The other states are Attention, Not restorable and No backup yet. Read more: How Restow proves restorability, Restore verification

Recycle bin (SharePoint and OneDrive) #

A holding area for deleted files. In Microsoft 365 for work or school, SharePoint keeps a deleted item for 93 days across the site recycle bin and the second-stage bin, and OneDrive keeps it 93 days unless an administrator changed the setting. Read more: Does Microsoft 365 back up your data?. Source: Microsoft Learn: SharePoint data deletion

Repository #

The place where backups are stored. In Restow a repository is a storage target, a local disk, S3-compatible object storage or an NFS share, with a role: primary, copy or previous. Read more: Storage

Restore check #

Restow's term Restow's term for its sample check: it draws a random sample from the newest backup (20 mails and 20 files plus a few calendar entries and contacts per mailbox, up to 20 files per machine or file share), reads it back through the restore path and compares it with the recorded checksums. It runs weekly for mailboxes, OneDrive and IMAP accounts, samples the library files of SharePoint sites the same way as a OneDrive, and runs after every new good backup for servers, clients and file shares; it feeds recovery readiness. Read more: How Restow proves restorability, How to prove that a backup can be restored

Restore point #

One stored state of a backup that can be restored. Snapshot and version are common synonyms, and the Restow interface uses all three. Read more: Grandfather-father-son backup rotation

Restore test #

Restoring from a backup on purpose, into a separate place, to prove that it works. The proof is only as wide as what was restored: a sample, one mailbox or a whole system. Restow's Proxmox VE boot test restores a guest and proves that it boots; a full test restore into a separate target for everything is planned. Read more: How to prove that a backup can be restored, How Restow proves restorability

Retention #

The rule for how long, and how densely, restore points are kept before they are pruned. It is not the same as a Microsoft Purview retention policy, which keeps or deletes content for compliance reasons. Restow can delay pruning on purpose: a restore point under an immutability lock stays until its lock ends, and a retention hold pauses retention for an object with an unresolved ransomware finding. Read more: Grandfather-father-son backup rotation, Does Microsoft 365 back up your data?, GDPR and backups

Retention hold #

Restow's term Restow's term for a pause of retention: while an object has a suspicious restore point that nobody has resolved, retention deletes no restore point of that object, so clean older restore points are not pruned while encrypted ones pile up. Only an administrator ends it, by marking the finding a false alarm or releasing the hold. It is not a legal hold, and backups go on, so storage grows. Read more: Ransomware detection, Grandfather-father-son backup rotation

Right to erasure (GDPR Art. 17) #

A person's right to have personal data erased without undue delay when a ground in Article 17(1) applies, for example that the data is no longer needed or that consent was withdrawn. Article 17(3) lists exceptions, among them a legal obligation to keep the data and the defence of legal claims. The Regulation does not mention backups; the EDPB took up erasure in backups in its 2026 report on the right to erasure. Read more: GDPR and backups, Backup vs. archive. Source: Regulation (EU) 2016/679, Article 17 (EUR-Lex)

RPO (recovery point objective) #

The point in time to which data must be recovered after an outage, which in practice sets how often you must back up. NIST defines it in SP 800-34 Rev. 1. Read more: RPO and RTO explained, The 3-2-1 backup rule. Source: NIST glossary: recovery point objective

RTO (recovery time objective) #

How long a system or its components can be in the recovery phase before the damage to the organization becomes unacceptable. Backups affect it through restore speed. NIST defines it in SP 800-34 Rev. 1. Read more: RPO and RTO explained, The 3-2-1 backup rule. Source: NIST glossary: recovery time objective

S

S3-compatible storage #

Object storage that speaks the Amazon S3 API, so a tool written for S3 also works with it. Providers differ in the optional parts, such as Object Lock, versioning and lifecycle rules, so compatibility is a matter of degree. Restow uses it as a repository, with a preset for Hetzner Object Storage; a new S3 endpoint must use HTTPS, except to localhost. Read more: Choosing backup storage, Storage

Scrub (storage check) #

A periodic read of stored backup data to find damage early. The Restow storage check reads a sample every week and every stored pack every month, and compares size and checksum with what was recorded. Since 0.4.2 it tells a pack it could not read for the moment (a connection or permission error) from a damaged one: the pack is read again and listed as temporarily unreadable, and only proven damage, or three checks in a row over at least 24 hours that could not read it, rates a backup as not restorable. Read more: Restore verification, How Restow proves restorability

Shared responsibility model #

The division of duties between a cloud provider and its customer. Microsoft documents that for Microsoft 365, as for other SaaS, the customer remains responsible for its data and identities while Microsoft runs the infrastructure. Read more: Does Microsoft 365 back up your data?. Source: Microsoft Learn: Shared responsibility in the cloud

SMB (Server Message Block) #

A network protocol for sharing files between computers, used by Windows file servers, by Samba on Linux and by most NAS devices. It exists in several versions, and which one a share speaks decides what a backup tool can use. Restow reads SMB 2.1, 3.0, 3.0.2 and 3.1.1 shares and refuses SMB1 and NTLMv1. Read more: File shares (SMB and NFS)

Snapshot #

A point-in-time view of data made by a storage system, a hypervisor or a backup tool. A snapshot that lives on the same storage as the original is not a backup by itself, because it shares its failure domain. Read more: The 3-2-1 backup rule

Synthetic full backup #

A full backup assembled from backup data already in the repository, the last full backup and the incrementals after it, without reading the source again. It keeps chains short without adding load to production systems. Read more: Full, incremental, differential and synthetic full backups compared. Source: Veeam: Backup methods

T

Tenant #

In Microsoft 365, one organization's directory and services (a Microsoft Entra tenant). In Restow, a customer organization inside an installation, with its own key, its own prefix in the repository and its own administrators; Service Provider supports many tenants in one installation. Read more: Multi-tenancy

V

Version history #

A SharePoint and OneDrive feature that keeps earlier versions of a file. The organization default for new libraries is a limit of 500 major versions with no expiry, and administrators can change it. It brings back an earlier state of a file, not a deleted site. Read more: Does Microsoft 365 back up your data?. Source: Microsoft Learn: Set default organization version limits

W

WORM (write once, read many) #

Storage that accepts data once and then refuses changes and deletion until a retention period ends. Amazon S3 Object Lock uses a WORM model; local disks and NFS shares have no hardware WORM, which Restow shows in its repository list, and the deletion delay it offers for them is not immutability. Read more: Immutable backups, WORM, air gaps and ransomware, The 3-2-1 backup rule. Source: Amazon S3 User Guide: Locking objects with Object Lock

Related articles

• The 3-2-1 backup rule Three copies, two kinds of storage, one location apart, and the extension with an immutable copy and zero errors. With a worked Microsoft 365 example.
• Grandfather-father-son (GFS) rotation Daily, weekly, monthly and yearly restore points with a calendar example, how it differs from keeping every backup for N days, and how it maps from tape to disk and object storage.
• Full, incremental, differential and synthetic full A comparison table for storage use, backup time, restore effort and chain risk, plus forever-incremental, changed block tracking and deduplication.
• Does Microsoft 365 back up your data? What Microsoft keeps, for how long (recycle bins, version history, retention), why retention is not a backup, and what Microsoft 365 Backup is.
• Backup vs. archive Recovery against retention, a comparison table, archiving vs journaling, why a rotating backup is not a record, the GoBD angle for German businesses, and how to run both.
• Immutable backups, WORM and air gaps How attackers go after backups, what Object Lock in governance and compliance mode does, what an offline copy adds, and what none of it covers.
• GDPR and backups What Article 17 asks of a backup, what the EDPB and German authorities say, why backups usually expire, and where locks and holds collide with an erasure request. Not legal advice.