# VPS Snaps > Automated server backups and cloud snapshots on eight clouds, kept in storage you own, with scheduled test restores, uptime monitoring and recovery servers. ## What is VPS Snaps? VPS Snaps is a SaaS platform that automates infrastructure backups for developers and agencies. It connects to your servers and cloud providers to run scheduled backups, storing everything to S3-compatible storage you control — VPS Snaps never hosts your backup data itself. ## Core Features - **VPS Snapshots** — Automated DigitalOcean, Hetzner, Vultr, Linode, AWS EC2, AWS Lightsail, Google Compute Engine, and Microsoft Azure server snapshots on a schedule. Azure is per managed disk rather than per machine: a VM with an OS disk and two data disks produces three snapshots per run, created and pruned as a set - **Database Backups** — PostgreSQL, MySQL, MariaDB and MongoDB backups over SSH (pg_dump / mysqldump / mongodump), plus managed Postgres on Railway, Render and Coolify - **Docker Backups** — Docker volume archives and docker-compose file backups over SSH, without stopping containers. Every named volume is required: one that cannot be archived fails the run. A volume holding SQLite databases is archived from the host with each database copied by SQLite's own backup API first (python3 or sqlite3 on the host), so apps such as Uptime Kuma and Vaultwarden keep every committed write - **File Backups** — Back up any file or directory path over SSH, with exclude patterns, as one full .tar.gz per run. SQLite databases under the chosen paths are copied with SQLite's own backup API first (python3 or sqlite3 on the server), so the archive has every committed write while the app keeps running; a restore removes stale -wal and -shm files - **Application Backups** — Scans a connected server for WordPress, Laravel, Craft CMS, Ghost, and Nextcloud, reading database credentials from each platform's own config, and creates matched File + Database jobs in one click - **Kubernetes Backups** — Namespace manifests plus best-effort persistent volume snapshots, via a service account token — no in-cluster agent: a short-lived Job reads the manifests inside the cluster and uploads them straight to storage, then is deleted; auto-detects and drives Velero if it's already installed - **DNS Zone Backups** — Every record in a Cloudflare zone, stored in the customer's own storage with the zone as a BIND file any DNS host imports, on the job's schedule and retention. Restored from the dashboard into the same zone: a preview lists every record changed, deleted or added since and writes nothing; the restore takes a safety copy of the zone, writes back only the records that differ, reads the zone back to confirm, and undoes what it wrote from the safety copy if Cloudflare refuses a change partway. Records added since are kept unless removal is chosen. Read through Cloudflare's API by the VPS Snaps worker (a zone has no customer machine to run on) and written straight to the customer's storage, never kept - **Shopify Store Backups** — A Shopify store backed up through Shopify's Admin API into the customer's own storage (any S3-compatible bucket, Google Drive, Dropbox, Box or OneDrive) on the job's schedule and retention, as one .tar.gz: products (variants, images, metafields), collections, customers, orders, draft orders, pages, blogs and posts, navigation menus, URL redirects, metaobjects, files (Content > Files), inventory levels, locations, discounts, markets, store settings and policies, and every theme's files. A per-job switch, "Include images and files" (on by default), also copies the image, video and file bytes rather than only their links; on Box and OneDrive, which hold a streamed archive in memory, the archive must stay under 64 MB. Restored from the dashboard into the same store: choose what to put back (a theme; products, all or by handle; collections; pages; blogs; menus; redirects; metaobjects; customers missing from the store), preview every difference without writing anything, then the restore takes a safety copy of the store and writes only what differs. A theme comes back as a .zip you upload in Shopify as an unpublished theme (or straight in as an unpublished copy where Shopify has given the app theme access), never over the live one. Nothing in the store is deleted except inside a restored product, which comes back exactly as backed up; existing customers are never changed. Not restorable: orders, draft orders, inventory levels, discounts and markets are kept as records only, because Shopify lets no app write an order back; gift cards are not restored either. Customer passwords are never in a backup or a restore (Shopify does not allow it), so restored customers sign in again. VPS Snaps' own Shopify app, which connects a store in one click, is awaiting Shopify's App Store review; until it is approved, a store connects through an app its organization makes in Shopify's Dev Dashboard (Client ID and secret; VPS Snaps gets a fresh 24-hour token for each backup), an app installed by link on a client's store (approved once by the store), or an Admin API token from a custom app made before 2026. Read by the VPS Snaps worker (a store has no customer machine to run on) and streamed straight to the customer's storage in memory, never kept. Free plan: one store, kept as its latest backup only, with images and files as links rather than copies; Starter plan and up: more stores, images and files copied, and the plan's longer retention - **WordPress Plugin** — The VPS Snaps plugin on wordpress.org (wordpress.org/plugins/vps-snaps) backs up a WordPress site's database and files from inside WordPress, whole multisite networks included, for hosts with no SSH. It uploads straight to the customer's own storage with short-lived signed links (Google Drive and S3 Object Lock buckets included, from plugin 1.0.2), and the site restores or moves from the dashboard. Every plan, Free included - **Agent** — For servers with no public SSH or API access. A Docker-based program the customer runs on the private machine holds a persistent outbound WebSocket connection to VPS Snaps (no inbound firewall rules or public IP needed) and executes Database, Docker, and File backups locally, uploading directly to the customer's own storage. Its image carries pg_dump, mysqldump and (from agent 1.2.6) mongodump and mongorestore - **Backup frequency by plan** — daily, every 12 hours, or every 5 minutes, depending on plan (see Pricing) - **Bring-your-own storage** — Amazon S3, Cloudflare R2, Backblaze B2, DigitalOcean Spaces, Wasabi, UpCloud Object Storage, any custom S3-compatible endpoint, the customer's own Google Drive (signed in with Google; VPS Snaps is granted only the files it creates, drive.file), the customer's own Box (signed in with Box; backups go to a folder in the customer's Box account, each within the Box plan's file-size limit, 250 MB on a free account), or the customer's own Dropbox (signed in with Dropbox; VPS Snaps is granted only its own app folder, Apps/, never the rest of the Dropbox), or the customer's own OneDrive, personal or work (signed in with Microsoft; VPS Snaps is granted only its own app folder, Apps/VPS Snaps, never the rest of the OneDrive); VPS Snaps sets no size limit of its own, and no storage is hosted or resold by VPS Snaps. Buckets with S3 Object Lock and a default retention are supported (AWS S3, Backblaze B2, Wasabi and other S3-compatible storage that supports it): every upload carries the checksum a locked bucket requires, VPS Snaps reads the lock and shows it on the Storage page, and never turns it on itself - **Hosts with no snapshot API** — Database, File and Docker jobs work over SSH on any server at any host, so a provider with no API (or one that cannot create a snapshot, such as SSD Nodes or InterServer) is still fully backupable — it just has no snapshot job type - **SSH Key Management** — Secure encrypted SSH key storage for server connections - **Restore** — Database, file and Docker backup runs, Kubernetes runs taken with Velero, WordPress sites (through the plugin) and whole applications (files and database together) can be restored from the dashboard onto any server in the workspace, not only the one the backup came from — including a new server that does not have the files yet, where the restore runs create-only and never overwrites an existing file. A run taken by the agent restores back onto the machine the agent runs on (agent 1.2.5 and later), with the same safety backup first. After a file restore every file it wrote is compared with the backup, byte for byte. A check-only mode downloads the archive to the target and verifies it against the checksum recorded at backup time without changing anything. A full restore takes a backup of whatever it is about to overwrite first and refuses to proceed if that fails. Database dumps carry DROP statements (pg_dump --clean --if-exists, mysqldump --add-drop-table) so they replace an existing database rather than only populating an empty one, and so a plain `gunzip -c dump.sql.gz | psql mydb` works by hand too. Limits: a restore returns the backup's contents rather than resetting the database exactly, so objects created after the backup survive; a MongoDB restore runs mongorestore --drop, so each collection in the archive replaces the one on the target while a collection created since the backup is left alone, and it is rehearsed with --dryRun first so a corrupt archive fails before anything is dropped; a Kubernetes run taken without Velero cannot be restored; provider snapshots are rebuilt from the provider's own console. DNS zone backups restore into their own Cloudflare zone (see DNS Zone Backups), and Shopify store backups into their own store, without orders (see Shopify Store Backups) - **Moving an Application to a New Server** — For a detected application (WordPress, Laravel, Craft CMS, Ghost): checks both servers read-only, creates the database and its user on the new server with the same credentials, takes a fresh backup of the database then the files, restores them onto the new server without overwriting any file already there, and counts files and tables on both sides. The old server is only ever read. Web server configuration, DNS, TLS certificates, cron and anything written after the backup are left to the operator. A database hosted elsewhere stays where it is and only the files move. Included on the Standard, Pro and Agency plans. See [Switching Providers](https://vpssnaps.com/switching-providers) - **Backup History** — Full run logs with success/failure status and file details - **Test Restores** — Opt-in per job: a throwaway server on the customer's own DigitalOcean, Hetzner, Vultr, Linode, AWS EC2, AWS Lightsail, Azure or Google Compute Engine account (a Railway or Render database is tested inside the customer's own project or workspace), the newest backup restored onto it, the data checked, the server destroyed, and an alert if it did not come back. File and Docker volume backups, and PostgreSQL, MySQL, MariaDB and MongoDB databases, are covered; provider snapshots, Kubernetes, WordPress plugin sites, DNS zones, Shopify stores, Agent jobs and Coolify databases are not. Every 30, 90 or 180 days per job, off by default because the throwaway server is billed to the customer's cloud account. Details at [Restore Testing](https://vpssnaps.com/restore-testing). - **Continuity Plans (disaster recovery planning)** — Pro and Agency: the systems a business must bring back, listed from what VPS Snaps sees, each with a recovery point and recovery time target; readiness checked against what the backups actually did (newest backup age, restores and test restores within the tier's window, measured restore times, recovery servers, copies kept with another company, failure alerts); one-click fixes where safe; a plan document with people, written procedures and runbooks, signed off as frozen versions with a yearly review reminder; and the evidence mapped to SOC 2 (A1.2, A1.3, CC7.5, CC9.1), ISO/IEC 27001:2022 (5.29, 5.30, 8.13, 8.14) and HIPAA 164.308(a)(7). It records evidence; it is not a certification, and VPS Snaps holds none and does not sign Business Associate Agreements. Pro includes one plan; Agency one per client. Details at [Disaster Recovery Planning](https://vpssnaps.com/disaster-recovery-planning). - **Recovery Servers** — A recovery plan per server (Standard, Pro and Agency plans). "Recover now" builds a replacement on the customer's own DigitalOcean, Hetzner, Vultr, Linode, AWS EC2, AWS Lightsail, Azure or Google Compute Engine account: from the server's newest snapshot when it has one (the whole machine), or otherwise a fresh Ubuntu with its databases (same names, users and passwords), files and Docker volumes restored from the newest backups, and the software the server ran — its packages (from the repositories it had added too), its web server, PHP, TLS certificate, cron and systemd configuration, its accounts and its services — recorded from every server with its backups while it was alive. When the server's own provider is down, it can be rebuilt from backups elsewhere: on a second cloud chosen in the plan ("Recover to "), on a cloud connected there and then, or on any Ubuntu 22.04 or 24.04 server the customer can reach as root at any host (checked first: root, supported Ubuntu, no database server already installed, disk space, outbound HTTPS; its host key shown to confirm; never deleted by VPS Snaps), with the provider that's down never asked anything; traffic then moves by Cloudflare DNS or by the records the summary lists, since a reserved IP can't cross providers. A readiness checklist says whether a server's backups are kept with a company other than the one running it. Its data is as old as the backups it was built from, and the page says how old before anything is built. A warm standby (Pro and Agency) keeps one built and refreshed from newer backups. Automatic failover (Pro and Agency; on DigitalOcean, Hetzner, Vultr, Linode, AWS EC2, Azure and Google Compute Engine by moving an address, and on any of them — AWS Lightsail included — by repointing the domain's DNS records on a connected Cloudflare account) checks the server every minute from VPS Snaps and, when that fails, from inside the customer's account too — from the warm standby when one is kept, otherwise from a small probe machine started beside the server for the outage and removed afterwards — and moves the server's reserved, floating, Elastic, static public or static external IP — or points the chosen Cloudflare records at it, proxied records at once — to the recovery server only after an outage both agree on; it never switches back on its own and stays off until re-armed - **Uptime Monitoring** — Paid plans (Starter and above). Monitors for web pages (status code, plus a word that must or must not appear), TCP ports, and scheduled jobs (heartbeats: the job calls a URL when it finishes, and is down when the call is late past its grace). An outage is confirmed by three failed checks 20 seconds apart, after the checker has confirmed its own connection, and held while many hosts fail at once; one alert when confirmed and one when back, to email, Slack, Microsoft Teams, Google Chat, Discord or (Agency) a webhook. A monitor linked to a server names that server's newest backups in its alert, with Recover now when it has a recovery plan and a link to rebuild it on another provider. Certificate expiry warnings at 14 and 3 days. Starter: 10 monitors every 5 minutes; Standard 30, Pro 100, Agency 300, every minute. Checks come from New York (165.227.255.101); no SMS or phone alerts yet. - **Status Pages** — Public pages at status.vpssnaps.com/ listing chosen monitors under chosen display names, with 90-day uptime bars, outages, and updates and planned-work notices the customer posts; never addresses or error text, and no analytics or ad scripts. Starter 1 page, Standard 3, Pro 10, Agency 50. On Agency, a page can be on the customer's own domain (one CNAME to domains.vpssnaps.com; Cloudflare issues the certificate) and without "Powered by VPS Snaps". - **Recovery Objectives** — Recovery point per job and per application against a schedule-derived target, and recovery time measured from real restores — shown as unmeasured until one exists, never estimated - **Notifications** — Email, Slack, Microsoft Teams, Google Chat, and Discord on backup success/failure, free on every plan; a custom arbitrary webhook URL is an Agency-plan feature - **Two-Factor Authentication** — TOTP 2FA with authenticator app + backup codes - **API Access and AI setup** — Standard plan and up. Bearer-token API keys, each either Read and run (list jobs and runs, start a backup) or Full access (also create agents, storage destinations, servers and backup jobs, with the same plan checks as the dashboard). The same key works on the REST API under /api/v1 and on an MCP server at https://vpssnaps.com/api/mcp. ChatGPT and claude.ai connect to the MCP server by OAuth instead of a key: the person signs in on a VPS Snaps page and chooses Read and run or Full access, and can disconnect it under Settings → API keys → Connected apps. So ChatGPT, Claude, Cursor and other MCP clients can set backups up: read the plan and storage, create an agent and its install command, create jobs, run them and read their logs. No MCP tool takes a storage key, cloud token or SSH key; storage is connected by a person in the dashboard. Editing, deleting and restoring are dashboard-only. Everything a key creates is in the workspace audit log. Details at [AI Setup](https://vpssnaps.com/ai-setup); instructions for an assistant at https://vpssnaps.com/ai-setup.md - **Public Status Page** — Live component health and incident history at /status ## Pricing Plans Monthly price, with the effective monthly price under annual billing (~33% off) in parentheses. All limits below are exact as configured today. - **Free** — $0/mo — 1 server, 1 backup job, daily backups, 1-day retention, 1 storage destination, 1 team member. No credit card required, ever. - **Starter** — $15/mo ($10/mo billed annually) — 5 servers, 3 backup jobs, daily backups, 3-day retention, 1 storage destination, 1 team member, 10 uptime monitors every 5 minutes, 1 status page - **Standard** — $45/mo ($30/mo billed annually) — 15 servers, 20 backup jobs, backups every 12 hours, 14-day retention, 3 storage destinations, 1 team member, database + Docker backups, moving sites to a new server, 30 uptime monitors every minute, 3 status pages - **Pro** — $85/mo ($57/mo billed annually) — 40 servers, 50 backup jobs, backups every 5 minutes, 30-day retention, 10 storage destinations, unlimited team members, database + Docker backups, moving sites to a new server, 100 uptime monitors every minute, 10 status pages, a continuity plan - **Agency** — $175/mo ($117/mo billed annually) — 150 servers, 250 backup jobs, backups every 5 minutes, 90-day retention, unlimited storage destinations, unlimited team members, database + Docker backups, moving sites to a new server, custom webhook notifications, 300 uptime monitors every minute, 50 status pages on your own domain, continuity plans (one per client) A feature-by-feature comparison against Snapshooter, a competing product, is at /compare/snapshooter. ## Why Choose VPS Snaps - **Cheaper at every matched tier than Snapshooter**, our closest competitor, at the same backup frequencies — full breakdown at /compare/snapshooter - **A genuinely free plan** — one server backed up daily for as long as it is used, no credit card and no trial clock - **Chat notifications on 4 platforms, not 1** — Slack, Microsoft Teams, Google Chat, and Discord, free on every plan including Free - **No vendor-hosted storage markup** — bring your own S3-compatible storage instead of paying metered per-GB fees to us - **Proof that backups restore** — scheduled test restores on a throwaway server in the customer's own cloud account, opt-in per job - **A way back when a server is gone** — recovery servers built on the customer's own cloud from the newest snapshot or backups (Standard and up), on another provider or any Ubuntu server when its own provider is down, with a warm standby and automatic failover on Pro and Agency - **Monitoring alongside backups** — uptime monitors, heartbeats and status pages on every paid plan, with the server's newest backup named in each outage alert ## Frequently Asked Questions **Do you store my backup data?** No. VPS Snaps stores only metadata (status, timestamps, checksums) — never the backup files themselves, which are kept in your own storage. Backup data takes one of three paths. Provider snapshots (and Kubernetes backups made with Velero) never leave your provider. Backups taken over SSH (files, databases, Docker volumes, Coolify databases) are uploaded by your server straight to your storage with short-lived signed links, as are those made by the VPS Snaps agent and the WordPress plugin, so that data never reaches us; if a server cannot upload for itself (no curl, or storage that refuses it), that run streams through our worker instead and its log says so. A Railway database, when Railway is connected with admin access, is dumped by a short-lived service in the customer's own Railway project, which uploads straight to their storage and is then deleted; a Render database is dumped by a job in the customer's own Render workspace, which does the same (Render needs a payment method on the workspace for it). Kubernetes manifests (without Velero) are read by a short-lived Job inside the cluster and uploaded straight to storage, where the cluster lets VPS Snaps run one. What streams through our worker by design: a Cloudflare DNS zone or a Shopify store, read through Cloudflare's or Shopify's API because neither has a machine of the customer's to run on (a Shopify store's data includes its own customers' names, emails and addresses); Kubernetes manifests where the cluster does not allow that Job, and managed Postgres on Railway connected without admin access, where pg_dump itself runs on our infrastructure; also a Railway or Render dump the platform will not run in the customer's account, which the run's log records. Restores and test restores do not: the server being restored downloads the backup straight from your storage with a short-lived link (falling back to our worker only if it cannot); a Railway or Render database is restored by a job in the customer's own project or workspace, which downloads it the same way (Render has no worker fallback). "Verify now" has the job's server download and check the backup, or asks your storage whether it is unchanged without downloading it. On the worker path the data is held in memory only — never written to a disk here and never retained. All three paths are set out in section 4 of the DPA at /dpa, and each job shows which one it takes. **Which VPS providers are supported?** DigitalOcean, Hetzner, Vultr, Linode, AWS EC2, AWS Lightsail, Google Compute Engine, and Microsoft Azure, all for automated snapshots. Database and Docker backups work over SSH on any server regardless of provider. Note: Linode is also known as Akamai Cloud Computing / Akamai Connected Cloud (Akamai acquired Linode in 2022) — same provider, same integration. Linode snapshots back up a single selected disk (typically the boot disk) rather than the whole instance, since Linode has no whole-instance snapshot API with independent retention. AWS EC2 snapshots create a full AMI (all attached volumes) in one region per connection. Google Compute Engine snapshots create a full machine image (all attached disks) in one project per connection. Microsoft Azure has no whole-machine equivalent at all — it snapshots managed disks individually, so VPS Snaps snapshots every disk attached to the VM and fails the run if any one of them fails, since a backup of only the OS disk would restore as a machine that boots with its data missing. Azure connects by signing in with Microsoft once, which gives VPS Snaps' own app Reader and a VPS Snaps role on the chosen subscription — snapshots, and the machines, disks and networking of recovery servers and test restores; not storage accounts or their keys, databases, key vaults, DNS or commands on machines — or with a service principal of the customer's own given the same role. It always runs as that app, never as the person, because a delegated token dies with a password change or the person who granted it leaving. One connection covers one subscription. Unmanaged (classic) VHD-backed VMs are refused by name. **What if my host isn't one of the supported providers?** Snapshot jobs need the host to expose an API call that creates a snapshot, and some hosts don't have one. SSD Nodes publishes fifteen API endpoints, two of which touch snapshots — one lists them, one restores one — and none creates one, so snapshots there are made by hand in the control panel and cannot be scheduled by any software. InterServer's published VPS API covers services, power state, a console screenshot and billing, with no disk or image operation. In both cases a snapshot integration is impossible rather than pending. Everything else still works: Database, File and Docker backups reach the machine over SSH and ask nothing of the host's API, so they run on any server at any host, including hosts with no API at all. See /ssd-nodes and /interserver. **How are managed database platforms backed up?** Railway, Render and Coolify hand you a Postgres database rather than a server, so there is nothing to SSH into. Railway connects with Railway sign-in (you choose which projects to share) or an account or workspace token (a project token will not work — different auth header). Signed in with admin access, the dump runs in a short-lived service in your own Railway project over its private network and uploads straight to your bucket, so the database needs no public networking and the data never reaches VPS Snaps. With a token, or less access, pg_dump runs on VPS Snaps infrastructure and streams to your bucket without being stored, and the database needs public networking enabled, since its DATABASE_URL points at .railway.internal which resolves only inside their network. Render connects with an API key. Each backup runs as a one-off job on a small on-demand service VPS Snaps keeps in your Render workspace (Render bills it at least $1 a month, and needs a payment method on the workspace), over Render's private network, uploading straight to your bucket. Without a payment method the dump runs on VPS Snaps infrastructure instead, streaming to your bucket without being stored, and the database must then accept external connections, which its Access Control setting can restrict to an IP allowlist. Coolify is different despite looking similar: it runs on a server you own, so its API is used only to locate the container and its credentials (the token needs the read:sensitive ability or passwords come back redacted) and the dump runs on your own host over SSH via docker exec, and that host uploads it straight to your storage. All three are Postgres only. **Do you use analytics or tracking?** Google Analytics runs on the public marketing pages only, and only for visitors who accept the consent banner — decline and no Google script is loaded and no Google cookie is written. There is no Meta Pixel and no advertising cookie. Analytics never runs inside the application: once signed in, how a customer uses VPS Snaps is not measured or sent anywhere. Signup attribution (which campaign or link brought someone) is captured first-party in VPS Snaps' own database, not by any analytics vendor. **Which storage providers are supported?** Any S3-compatible storage: Amazon S3, Cloudflare R2, Backblaze B2, DigitalOcean Spaces, Wasabi, UpCloud Object Storage, or a custom S3-compatible endpoint. Or Google Drive: backups go to a folder in the customer's own Drive, uploaded there directly by their servers, agents and WordPress sites; restores download straight from it with a token that lasts under an hour; retention deletes old backups permanently, freeing the space. Drive counts backups against the Google account's storage and limits uploads to 750 GB a day. Or Box: backups go to a folder in the customer's own Box account. Box takes each backup as one finished file, so the server, agent, WordPress site or dump container writes it to its own disk first and needs free disk about the size of the backup; it uploads with a token Box limits to that job's folder for about an hour, and downloads use Box's own download link (about 15 minutes), never a token. The Box plan sets the largest file: 250 MB on a free personal account, more on paid plans (5 GB on Personal Pro and Business); VPS Snaps reads the account's limit and refuses a larger backup before uploading. Backups that stream through VPS Snaps reach Box only up to 64 MB, because VPS Snaps never writes customer data to its own disk. Box has no narrower per-app permission, so the grant covers all files and folders; VPS Snaps keeps to the destination's own folder. Retention deletes old backups and purges them from Box's trash, freeing the space. Or Dropbox: the Dropbox app has App folder access only, so VPS Snaps can see and write only its own folder, Apps/, in the customer's Dropbox, and backups go to a folder the customer names inside it. Dropbox takes uploads as they stream, so no spare disk is needed and backups that stream through VPS Snaps have no 64 MB limit (they pass through in memory, as with S3 and Google Drive). Dropbox accepts files up to about 2 TB through its API; the customer's Dropbox space is the limit. Each backup is checked against Dropbox's content hash, computed as it uploads. To upload, a server, agent, WordPress site or dump container gets the destination's own short-lived access token (it reaches only the app folder and lasts at most 4 hours), or, for a file up to 150 MB from an agent or site, a one-time upload link with no token; downloads use Dropbox's temporary link (4 hours), never a token. Retention deletes old backups; Dropbox's own deleted-files history applies (Dropbox keeps deleted files restorable by the customer for 30 to 365 days, depending on their plan, and deleted files do not count against their space). Or OneDrive, with a personal Microsoft account or a work or school one: VPS Snaps has app folder access only (Files.ReadWrite.AppFolder), so it can see and write only Apps/VPS Snaps in the customer's OneDrive, and backups go to a folder the customer names inside it. OneDrive takes each backup as one finished file, so the server, agent, WordPress site or dump container writes it to its own disk first and needs free disk about the size of the backup; it uploads to an upload session VPS Snaps opens for that one file, and no Microsoft token is handed out. Each backup is checked against OneDrive's quickXorHash. Downloads use OneDrive's own short-lived download link. A single file can be up to 250 GB; the customer's OneDrive space is the limit. Backups that stream through VPS Snaps reach OneDrive only up to 64 MB, because VPS Snaps never writes customer data to its own disk. Retention deletes old backups permanently where the account allows it, otherwise to OneDrive's recycle bin. Work accounts may need their Microsoft 365 administrator to approve the app first. The agent needs 1.4.0 or later for OneDrive, the WordPress plugin 1.2.0 or later. **How does database backup work?** VPS Snaps connects over SSH and runs pg_dump (PostgreSQL), mysqldump (MySQL/MariaDB) or mongodump (MongoDB) on your server, and the server uploads the compressed output straight to your storage destination with short-lived signed links, so it never passes through us. The VPS Snaps agent can run the same dumps with no inbound SSH at all. The SQL engines produce `-.sql.gz`; MongoDB produces `-.archive.gz`, a BSON container rather than SQL. mongodump ships in mongodb-database-tools, packaged separately from the MongoDB server, so the job checks it is installed before running rather than uploading an empty archive. **How does Docker backup work?** VPS Snaps connects over SSH and archives your Docker named volumes and docker-compose files via a temporary container, then uploads the compressed archive to your storage destination. **How does file backup work?** VPS Snaps connects over SSH, archives the file or directory paths you configure (skipping any that don't exist rather than failing the whole run), and uploads the compressed archive to your storage destination. **How does application backup detection work?** From a connected server's detail page, clicking "Detect applications" scans common web-root locations (/var/www, /home/*/public_html, Cloudways' and Forge's conventions, and more) for WordPress, Laravel, Craft CMS, Ghost, and Nextcloud. Database credentials are read from each platform's own config file (wp-config.php, .env, config.production.json, Nextcloud's config/config.php) where the format allows reliable parsing. One click per detected app creates a matched File + Database job pair, linked as one application: the database is backed up first and the files right after it, never at the same time, and the two can be restored together in one step — including onto a new server. Other applications are still detected and file-backed-up, just without automatic database credential extraction. The pair is taken in sequence, minutes apart, not as a single frozen moment: the application keeps running during the backup. **How does Kubernetes backup work?** VPS Snaps connects to your cluster's Kubernetes API using a service account token. If Velero is already running in the cluster, VPS Snaps detects it and drives it directly for a fuller backup. Otherwise, a short-lived Job in the cluster, running as that service account, reads namespace manifests (ConfigMaps, Secrets, Deployments, StatefulSets, and more) and uploads them as gzipped JSON straight to your own storage destination (without the setup guide's Job grant, the worker reads them over the API instead) and attempts native volume snapshots for any PersistentVolumeClaim whose StorageClass supports CSI snapshots, skipping and logging the rest. **How does the Agent work for private infrastructure?** The Agent is for servers with no public SSH or API access — behind a NAT gateway, a VPN, or deep inside a private VPC. Instead of VPS Snaps connecting in, a small Docker-based program run on the server connects out and holds a persistent WebSocket connection, waiting for job instructions. It's the same Database, Docker, and File job types VPS Snaps already supports — Agent is an alternative connection method to Server (SSH), not a separate feature. The agent uploads backups directly to the workspace's own storage destination; VPS Snaps never sees the data during the backup, only job status and logs. ("Verify now" on an agent backup asks the storage whether it is unchanged, without downloading it.) **What happens if a backup fails?** You're notified immediately via email, Slack, Microsoft Teams, Google Chat, or Discord (whichever you've configured), and the failure is logged with full details in your backup history. **How do I restore a backup?** From the dashboard, for database, file and Docker runs, Kubernetes runs taken with Velero, and WordPress sites through the plugin: open a successful run, choose the target, and confirm by typing the server's name. A run taken by the agent restores onto the agent's own machine the same way. A detected application's files and database restore together in one step. A check-only mode verifies the archive on the target without changing anything, and a new server without the files can be restored onto in create-only mode, which never overwrites an existing file. Provider snapshots are restored from the provider's console. A DNS zone or a Shopify store restores into the zone or store it came from, after a preview and a safety copy; a Shopify store's orders cannot be written back. Every artefact is an ordinary .sql.gz or .tar.gz in the customer's own storage, so manual restore always remains possible without VPS Snaps. **How do I know a backup would actually restore?** Scheduled test restores, opt-in per job: VPS Snaps creates a throwaway server on the customer's own DigitalOcean, Hetzner, Vultr, Linode, AWS EC2, AWS Lightsail, Azure or Google Compute Engine account, restores the newest backup onto it through the ordinary restore pipeline, checks the data arrived (tables or collections in a database created empty, files under the job's paths or inside its Docker volumes), alerts if it did not, and destroys the server. A test that could not run at all (no connection to build on, a server that could not be created or set up) is not reported as a failed restore: it is retried within hours, and after three attempts in a row the customer is told the backup has not been tested. Covers file and Docker volume backups, and PostgreSQL, MySQL, MariaDB and MongoDB. Railway and Render Postgres backups are tested in the customer's own Railway project or Render workspace instead: a throwaway Postgres is created there, the backup restored into it, the tables counted, and the database deleted. Every backup is also validated as it uploads — the archive must decompress, tar headers must be intact, database dumps must reach their completion marker — and read back from the bucket to confirm its size. **What does retention delete?** Everything older than the job's window (capped by plan): provider snapshots through the provider's API, and dumps and archives from the customer's bucket. Never a job's newest backup, anything a restore is reading, or the safety copy of a restore that did not succeed; and only objects VPS Snaps wrote, by the exact key recorded at upload — never by listing or prefix. **Can I move a site to a new server or another host?** Yes, for an application VPS Snaps has detected. It checks both servers first, creates the site's database and user on the new server, takes a fresh backup of the database then the files, restores them there without overwriting any file already present, and compares the counts. The old server is only read, so a failed move leaves the live site unchanged. DNS, the web server configuration, HTTPS certificates, cron, and writes made to the old site after its backup are the operator's to handle. Included on the Standard, Pro and Agency plans. Details at [Switching Providers](https://vpssnaps.com/switching-providers). **Can I run backups manually?** Yes — any backup job can be triggered on demand from the dashboard, in addition to its schedule. **Is there a free plan?** Yes — one server backed up daily, no credit card required and no trial to expire. It lasts as long as it is used: a free account with no sign-in and no successful backup for 60 days is deleted, after the owner is emailed at 14, 30 and 45 days. Paid accounts are never deleted for inactivity. The free plan is one account per person or organisation, and a second free account cannot add a server, cloud account, storage account or site another free account already uses. **How do I delete my account?** In the dashboard, under Settings → Delete account, by typing the account's email address and password (a paid plan is cancelled under Billing first). The account, its workspace, servers' settings, jobs and history are deleted from VPS Snaps, and the sign-ins given to VPS Snaps are revoked where the provider allows it. Backups in the customer's own storage, and anything on their servers and cloud accounts, are not touched. **Can I cancel anytime?** Yes. Cancel anytime from the billing page; your subscription stays active through the end of the paid period, no long-term contract. ## Pages - [Home](https://vpssnaps.com/): What VPS Snaps does, and who it is for - [Features](https://vpssnaps.com/features): Every capability that ships, grouped by category, and where each one stops - [Pricing](https://vpssnaps.com/pricing): Plans, what counts against each limit, and what happens on upgrade, downgrade or cancellation - [VPS Snapshots](https://vpssnaps.com/vps-snapshots): Provider-native snapshots across eight clouds, created inside the customer's own account - [Database Backups](https://vpssnaps.com/database-backups): pg_dump, mysqldump and mongodump over SSH, plus managed Postgres on Railway, Render and Coolify - [Docker Backups](https://vpssnaps.com/docker-backups): Named volumes archived by a throwaway helper container, without stopping anything - [File Backups](https://vpssnaps.com/file-backups): Any path on the server, one gzipped tar per run, with exclude patterns - [Application Backups](https://vpssnaps.com/application-backups): Scans a server for WordPress, Laravel, Craft CMS, Ghost and Nextcloud and creates matched file and database jobs - [Switching Providers](https://vpssnaps.com/switching-providers): Moving a detected application to a new server, what is checked first, and what the operator still does - [Kubernetes Backups](https://vpssnaps.com/kubernetes): Velero when it is installed, otherwise namespace manifests and CSI volume snapshots - [Uptime Monitoring](https://vpssnaps.com/uptime-monitoring): Monitors for sites, ports and scheduled jobs that confirm an outage before alerting and put the server's newest backup and Recover now in the alert, plus status pages, on the customer's own domain on Agency - [Restore Testing](https://vpssnaps.com/restore-testing): Scheduled test restores of the newest backup onto a throwaway server on the customer's own cloud, what counts as passing, and what is not covered - [Disaster Recovery Planning](https://vpssnaps.com/disaster-recovery-planning): Continuity plans built from the customer's backups, readiness checked against real restores, a signed-off document, and evidence mapped to SOC 2, ISO 27001 and HIPAA - [DNS Backups](https://vpssnaps.com/dns-backups): Every record in a Cloudflare zone kept in the customer's own storage with a BIND file, and restores that preview every difference and write back only those - [Shopify Backups](https://vpssnaps.com/shopify-backups): A Shopify store's products, customers, orders, content and every theme kept in the customer's own storage, and restores that preview every difference, hand back any theme as a .zip to upload, and keep orders as records - [Backup Agent](https://vpssnaps.com/agent): An outbound-only Docker container for machines with no inbound route - [AI Setup](https://vpssnaps.com/ai-setup): Setting backups up from ChatGPT, Claude Code, Cursor or another MCP client, what the assistant can and cannot do, and the REST API - [Cloud Providers](https://vpssnaps.com/providers): How each snapshot integration works, and what to do when a host has no snapshot API - [Cloudways](https://vpssnaps.com/cloudways): Per-application backups over the SSH that Cloudways exposes under its managed layer - [SSD Nodes](https://vpssnaps.com/ssd-nodes): Why no snapshot integration is possible there, and what runs over SSH instead - [InterServer](https://vpssnaps.com/interserver): Per-site backups on cPanel and DirectAdmin boxes over SSH - [vs SnapShooter](https://vpssnaps.com/compare/snapshooter): Near-identical tier shapes at lower prices; their plans include storage and list encryption and region replication, ours add test restores, uptime monitoring and recovery servers; and where SnapShooter wins - [vs SimpleBackups](https://vpssnaps.com/compare/simplebackups): Broader coverage at roughly three times the price, and where each one wins - [vs Ottomatik](https://vpssnaps.com/compare/ottomatik): The same job counts per tier; theirs are database and file jobs with storage included, ours add snapshots, Docker, Kubernetes and test restores - [vs cron scripts](https://vpssnaps.com/compare/cron-scripts): A crontab line and a pg_dump script against VPS Snaps: five ways scripts fail without saying so, when to keep one, and how to move off it - [vs Weap.io](https://vpssnaps.com/compare/weap): Weap.io costs less per server and snapshots ten clouds; VPS Snaps adds dashboard restores, test restores and more backup types - [Learn](https://vpssnaps.com/learn): Tested, provider-neutral tutorials: snapshots on eight clouds; PostgreSQL, MySQL, MongoDB, Redis and SQLite backups, restoring PostgreSQL dumps and importing MySQL SQL files, repairing MySQL tables, resetting the MySQL root password, backing up MySQL replicas and MongoDB replica sets, managed databases, Railway and Render Postgres, point-in-time recovery, PostgreSQL incremental backups, pgBackRest, streaming database backups to S3, XtraBackup, PostgreSQL major upgrades, Elasticsearch/OpenSearch snapshots, ClickHouse, InfluxDB, CouchDB, Neo4j, Qdrant and Meilisearch; S3/R2/B2/Wasabi/Spaces/UpCloud/Hetzner/Scaleway/OVHcloud/Vultr/Akamai/IDrive e2/Google Cloud Storage/Azure Blob/Google Drive storage setup, Hetzner Storage Boxes and copying a bucket to another provider; Docker volumes and images, Docker Compose databases, Coolify, Dokku, Portainer, Kubernetes with Velero and etcd, k3s; WordPress (including moving it to a new host), WooCommerce, Shopify stores, Joomla, Drupal, Magento, PrestaShop, Moodle, Laravel, Ghost, Craft CMS, Nextcloud, Vaultwarden, n8n, Gitea, GitLab, Discourse, Mattermost, Keycloak, Grafana, Odoo, Matomo, Immich, Mastodon, Uptime Kuma, Plausible Analytics, Paperless-ngx and Minecraft servers; tar (including incremental backups), rsync, rsnapshot, etckeeper, scp, SSH key login, rclone, restic, Borg, Kopia (and how the three compare), Duplicati, GPG encryption, full disk images with dd, LVM, ZFS and Btrfs snapshots, Proxmox VE, cron and systemd timers, find, backing up before an Ubuntu release upgrade, a server backup checklist, cPanel, Contabo, Hostinger, OVHcloud, Oracle Cloud and Scaleway servers and moving to a new provider; business continuity vs disaster recovery, RPO/RTO, retention, ransomware, Cloudflare DNS, DNS failover, rebuilding a server, recovering a hacked server, recovering deleted files, a DR plan template, the 3-2-1 rule and restore testing; uptime monitoring, status pages and backup failure alerts - [Help Center](https://vpssnaps.com/help): Setup guides and troubleshooting, from first SSH connection to diagnosing a failed run - [Blog](https://vpssnaps.com/blog): Guides on backing up servers, databases, Docker and Kubernetes - [FAQ](https://vpssnaps.com/faq): The questions that come up before signing up, and the ones that come up at 3am - [About](https://vpssnaps.com/about): Who builds and runs VPS Snaps (Scott Matarazzo, founder), and how it treats customer data - [Changelog](https://vpssnaps.com/changelog): New features since the first version in June 2026, newest first, each dated the day it went live - [Contact](https://vpssnaps.com/contact): How to reach support, and what to include in a backup-failure report - [Brand](https://vpssnaps.com/brand): The logo files, colors with hex codes, typography, social banners and descriptions, for anyone writing about VPS Snaps - [Status](https://vpssnaps.com/status): Live component health and incident history - [Security](https://vpssnaps.com/security): Encryption at rest, two-factor authentication, and how credentials are handled - [Trust Center](https://vpssnaps.com/trust): Every security commitment in one place: subprocessors, status, compliance progress (working towards SOC 2 and ISO 27001), and reporting a vulnerability ## Provider Guides - [DigitalOcean Backups](https://vpssnaps.com/providers/digitalocean): Whole-Droplet images, created and retained in the customer's own account - [Hetzner Cloud Backups](https://vpssnaps.com/providers/hetzner): API token scoped to a single project - [Vultr Backups](https://vpssnaps.com/providers/vultr): API keys, with Vultr's IP allowlisting accounted for at setup - [Linode Backups](https://vpssnaps.com/providers/linode): The Images API, because Linode has no whole-instance snapshot API — one image per disk - [AWS EC2 Backups](https://vpssnaps.com/providers/aws-ec2): AMIs, with the EBS snapshots behind them managed and deleted on deregistration - [AWS Lightsail Backups](https://vpssnaps.com/providers/aws-lightsail): Instance snapshots of the whole instance, attached disks included, one region per connection - [Google Compute Engine Backups](https://vpssnaps.com/providers/google-compute-engine): Machine images, capturing every attached disk in one resource; connected by Google sign-in or keylessly (Workload Identity Federation, no key) - [Microsoft Azure Backups](https://vpssnaps.com/providers/azure): One snapshot per attached managed disk, created and pruned as a set ## Legal and Policy - [Terms of Service](https://vpssnaps.com/terms): Access, acceptable use, billing and termination - [Privacy Policy](https://vpssnaps.com/privacy): What is collected, how credentials are encrypted, and which third parties are involved - [Refund and Cancellation Policy](https://vpssnaps.com/refund-policy): Fourteen-day money-back guarantee, and how plan changes are billed - [Cookie Policy](https://vpssnaps.com/cookies): Every cookie set, what it does, and how long it lasts - [Data Processing Addendum](https://vpssnaps.com/dpa): GDPR roles, subprocessors, security measures and deletion on termination ## Feeds and Machine-Readable - [Blog RSS feed](https://vpssnaps.com/blog/rss.xml): The 20 most recent articles, newest first - [Sitemap](https://vpssnaps.com/sitemap.xml): Every public URL - [Blog sitemap](https://vpssnaps.com/blog/sitemap.xml): Article URLs, generated from the live article list ## Account - [Create an account](https://vpssnaps.com/register): The Free plan needs no credit card - [Sign in](https://vpssnaps.com/login) ## Company VPS Snaps is built and operated by BrokeHold. - [Support](mailto:support@vpssnaps.com): Backup failures, setup and billing - [Privacy and data requests](mailto:privacy@vpssnaps.com)