Backups fail silently
A full disk, a folder someone moved, an expired password. The job is still scheduled, but it has backed up nothing for weeks and nobody noticed.
Managed backup
An agent on every server, encryption before anything leaves your company, storage away from your premises and a panel that warns you when something breaks. Without opening a single port on your network.
Almost nobody finds out their backup was useless on a good day. They find out on the day of the disaster.
A full disk, a folder someone moved, an expired password. The job is still scheduled, but it has backed up nothing for weeks and nobody noticed.
Having backups is not the same as being able to recover. A corrupted file or a wrong setting only shows up when you need it, with the business stopped.
Modern ransomware looks for the backups first and destroys them before encrypting the data. If the backups live on the same network, they are lost along with everything else.
Three pieces, with nothing for you to administer.
A small service on each server, installed with a single line. It updates itself and needs no maintenance. Your network opens no ports: the agent is the one that calls out.
Data is compressed, deduplicated and encrypted inside your company, before it leaves. What travels and what is stored is unreadable without the key, which is not kept in the storage.
The result is kept in professional storage in a European data centre. The panel checks every backup, every repository and every server, and warns when something does not add up.
Files and databases, each with the right method.
Any folder on the server, with exclusions for what is not worth copying.
A consistent copy of each database without interfering with your maintenance plan. Optionally, a transaction-log backup every few minutes so you can recover to a precise time of day.
Logical dump using the engine's official tools.
Logical dump using the engine's official tools.
Microsoft keeps deleted items for 93 days and no longer. After that, if nobody backed it up, there is nowhere to get it from.
The same engine with another provider: personal drives, shared drives and team spaces.
Messages are stored in their original format, one per file, so they can be opened years from now without depending on Microsoft.
Nothing installed on the NAS: a server that already has the agent backs it up.
The only thing that matters about a backup is being able to go back.
Every backup keeps the complete state at that moment. You browse the contents like a folder, pick a file, a folder or everything, and recover it to the destination you choose.
A recovery never writes over the original data or touches the operating system: it is placed aside and you decide what to move back. Databases are always restored under a new name, never over the one in production.
The history of every recovery is kept, with who requested it, what was recovered and how it ended.
When a file is needed and nobody remembers which month it was from, it is searched by name across every backup at once, not one by one.
Once a quarter a drill is run: a full backup is really restored to another location, timed. That produces the two numbers that matter in a contract, how long it takes to get back to work and how old the recovered data is, and they are put in writing.
Every week the system picks a random file from each server's latest backup, recovers it and checks with a cryptographic fingerprint that it is identical to the original. If anything is off, an alert fires. Nobody has to remember to test it: the result appears in your monthly report.
31 automatic checks, each with its own threshold and recipients.
Alerts arrive by e-mail, by Telegram or into your own system through a signed webhook, and they are always recorded in the panel.
On top of that, a single message every morning: servers offline, backups that failed overnight, open alerts and machines that need maintenance. If all went well, one line.
The design starts from an uncomfortable idea: one day one of your servers may be compromised. The backups have to survive that.
Data is encrypted on your own server. Neither the storage provider nor anyone with access to the disks can read it.
It only performs a closed set of foreseen operations. There is no way to tell it to run any command, not even from the panel.
Deletion by age is always ordered by the panel. If backups disappear from a repository without being asked for, a critical alert fires: that is exactly what ransomware would do.
The storage also keeps its own snapshots, out of reach of your servers' credentials. Even if someone wiped everything, the previous snapshot is still there.
The agent only accepts cryptographically signed versions, verifies them before installing and rolls back on its own if the new one does not start.
Who signed in, from which address, what changed, what was recovered. With export for audits.
When signs of mass encryption are detected, age-based cleanup is frozen for a month on that server. That is what stops the good backups from expiring exactly while the incident is being investigated.
Once a day the panel looks at the storage itself and compares what is there with what the server reports. If a server were compromised and kept saying everything was fine, that is where it would show.
Each server's credentials are limited to its own folder: a compromised server cannot read or delete another one's backups.
Transparency without having to ask for it.
Every month (or every week) you receive a PDF with the backups made, the compliance percentage, the incidents, the space used and the result of the recovery tests.
The report includes compliance with the target of one correct backup per server per day, expressed as a percentage. No need to rely on an impression.
You can have your own read-only user to sign in whenever you like and see your servers, your backups, your calendar and your recoveries. Without seeing anything from other clients or touching any setting.
A document with every recovery check in the period, the file checked, its cryptographic hash and the result of the drills. It is what an audit or a cyber-risk insurer asks for, and it carries its own recorded hash.
The GDPR article 28 annex, with where the data is, how long it is kept, who else is involved and what happens when the service ends. Generated from the real configuration, not from a template.
For anyone who wants the small print.
| Source types | Folders and files, SQL Server, MySQL, MariaDB, PostgreSQL, OneDrive and SharePoint, Google Drive, Dropbox, Microsoft 365 mailboxes and network folders |
|---|---|
| Technology | restic over SFTP storage: deduplication, compression and authenticated encryption at source |
| Scheduling | Flexible per job, in local time, with automatic catch-up of missed backups if the server was off |
| Retention | By number of recent, daily, weekly and monthly copies, configurable per job |
| Bandwidth | Upload limit in MiB/s within a time window, so the line is not saturated during working hours |
| Open files | Volume snapshot (VSS) on Windows |
| Manual run | Launch a backup immediately, with live progress and the option to cancel |
| Logging | Full log of every run, exactly as the backup engine returns it |
| Browser | Navigate the contents of each backup, folder by folder |
|---|---|
| Selective recovery | Files, folders or the whole backup, to the destination you choose |
| Protection | Never over the original location, nor over system folders, nor to network drives |
| Databases | Restore as a new database, never over the existing one |
| Automatic test | A random file is recovered every week and checked against a SHA-256 fingerprint; the interval is set from the panel |
| Comparison | Differences between two backups: what was added, removed or changed |
| History | Record of every recovery, exportable |
| Search by company | Find a file across every server of a company at once, without knowing which one it was on |
| Direct download | Fetch the file to the panel and download it from the browser, with no need for the original server to work |
| Which date to go back to | Each backup is marked clean, doubtful or tainted based on how the data behaved that night |
| Disappearances | A dated record of folders that stopped being there, to answer «since when has this been missing?» |
| Right to erasure | Remove a path from every backup in the history, with a dry run first and a record of what was removed |
| Live status | A heartbeat every minute per server, and every ten seconds while someone has that server open; the panel refreshes itself |
|---|---|
| Rules | 31 checks with configurable threshold, severity and channels |
| Exceptions | Different thresholds for one server or for a whole company, with the reason in writing |
| Periodic review | Chronic failures, misconfigured jobs, servers that no longer fit in their night window and real gaps without a backup |
| Notification drill | An end-to-end test that the alerting chain works, with how long it took and which channels it went out on |
| Channels | E-mail (SMTP, Gmail or Microsoft 365), Telegram and signed webhooks |
| Daily digest | One message a day with whatever needs attention |
| Maintenance windows | Mute alerts for a server or the whole company for a while, with a reason |
| Calendar | Month and week view with what was done and what is planned |
| Timeline | Runs per server over 24 hours or 7 days, with overlap detection |
| Updates | Pending system updates, how many are security ones and whether a reboot is pending |
|---|---|
| Performance | CPU and memory with 7 days of history |
| Inventory | Hardware, OS, physical disks with S.M.A.R.T. status, database instances, services, IP addresses |
| Change history | A record of every hardware, version or service change |
| Read only | The agent reports; it never installs updates or modifies the system |
| Reports | Monthly or weekly PDF per company, automatic by e-mail, with the provider's branding |
|---|---|
| Compliance | Percentage of days with a correct backup per server |
| Space forecast | Estimate of when the contracted space will run out |
| Exports | Runs, recoveries and activity log as CSV |
| Client access | Read-only user limited to their own company |
| Languages | Spanish, Catalan and English, in the reports and alerts too |
| Incident case file | One button freezes retention across the whole company, opens the case file and counts the 72-hour article 33 deadline |
| Cyber-insurance sheet | The controls policies ask about, answered with real, dated data, including the ones that are not met |
| Recovery sheet | A PDF per server with what is backed up, how much history there is and who to call, meant to be printed and kept off-site |
| Documents with a hash | Every report issued is recorded: paste its hash and you can confirm it came from here |
| Extra line items | Restores, hours and onboarding are written down and go into the month's amount |
| Encryption | At source, with a key that does not live in the storage |
|---|---|
| Isolation | Own credentials and repository per server |
| Agent | Closed set of operations; no arbitrary command execution |
| Updates | ECDSA-signed packages, verified before installing, with automatic rollback |
| Storage identity | The storage server's key is pinned and verified on every connection |
| Panel access | Two-factor verification, account lockout on failed attempts and origin blocking against password spraying, password recovery by e-mail and separate roles |
| Audit | Complete record of access and changes, with IP address |
| Cloned server | Two servers with the same identity are detected and the second is blocked, so they do not mix their history |
| Scoped keys | Key export can be limited to one server or one company, and exactly what was taken is recorded |
| Clean retirement | The agent uninstalls itself from the panel when you finish with a client, without logging into their server |
| Full removal | When you are done, a company is deleted in full: its data in the panel, its backups in storage and its access credentials, with a record of what was deleted |
Very little, and nothing that changes how your servers work.
In professional storage in a European data centre, in a folder exclusive to each server and with its own credentials. Data leaves your company already encrypted.
Not without the encryption key, which is not kept in the storage. Neither the storage provider nor anyone with access to those disks can see the contents.
Backups are not deleted from the server: deletion by age is ordered by the panel, and a disappearance without an order triggers a critical alert. The storage also keeps its own snapshots out of reach of your servers' credentials.
Because it tests itself. Every week a random file is recovered and checked to be identical to the original, and the result appears in your report. The frequency is whatever you want it to be, down to a daily test.
The backup runs at the agreed time, usually overnight, and the upload speed can be limited so the line is not saturated during working hours.
The backup is recorded as missed and is caught up automatically when the server comes back, within the configured window.
Yes. You can have read-only access to your own company, on top of the periodic report by e-mail.
Your backups are yours. The encryption keys and the data are handed over, and they can be read with standard open-source tooling, without depending on this panel.
We will explain how it would look in your particular case, which servers are worth backing up and how often.