Sooner or later something gives way: a drive dies, a storm takes the mains out, a digger in the street slices the fibre, someone in a hurry opens a booby-trapped attachment. What really counts is how soon you can raise invoices and take orders again. We therefore start from three figures: how long each application may sit idle, how many hours of typing you are prepared to redo, and who makes the decisions during the first half hour. Backups, alerting and spare capacity are then sized around those answers, never the reverse.
Only figures go into the first one; the remaining five take their dimensions from it. You never fund resilience your activity cannot justify, and every line of the quotation has a stated reason.
For the ERP, the mailboxes, the web shop and the file server we write down two values: the outage you can live with (RTO) and the amount of work you accept losing (RPO). That single page then governs every technical decision taken afterwards.
A written schedule, version history long enough to cover a mistake noticed late, one copy away from the building and one immutable copy that nobody can wipe or encrypt. At the frequency agreed in the contract a sample is pulled back into a sealed-off environment and checked.
Servers, Microsoft 365, carrier links, remaining disk space, certificate expiry dates, business services that quietly fail to restart. On the Premium plan the alerts are worked day and night, so the fault is usually cleared before the first user picks up a phone.
A second internet line taken out with a different carrier, switching over on its own, and for the systems that deserve it a replica held at a European hosting provider that picks up the work once the main server falls silent.
The UPS is configured to bring the servers down cleanly as soon as mains power vanishes, and we watch its batteries age. When a drive or a battery has to be swapped, we walk your colleague or the local contractor you appoint through it over a video call.
A short document for the day everything stops: the order in which applications come back, who does what, the list of people to contact, the paper fallback and a standard message for customers while the service is still missing.
Five thoroughly ordinary scenarios. Every one of them calls for a separate safeguard, and no safeguard stands in for another.
Answer: backup plus replica. The commonest case, and the easiest to prepare for. Swapping the hardware belongs to the manufacturer's support or to your own staff; putting the data and the system back into service belongs to us, remotely.
Answer: a copy beyond reach. Today's attacks hunt down the backup repositories first and scramble those too. That is the reason for the immutable copy, which keeps separate credentials and multi-factor sign-in, independent of your directory.
Answer: a long history. A contracts folder gone from SharePoint tends to surface weeks afterwards, when a dispute makes somebody look for it. A backup reaching no further than yesterday is then worth nothing.
Answer: a second carrier. Two lines bought from one supplier frequently share a duct and drop in the same minute. In plenty of offices a 4G or 5G router is enough to keep going while the repair crew works.
Answer: a UPS and a controlled shutdown. An abrupt loss of mains can harm the database behind your accounting package far more lastingly than the interruption itself, and the repair is then counted in days.
Nothing counts as a backup until somebody has brought it back. The unwelcome discoveries are always the same ones: a job that has been failing for months with nobody reading the report, a folder the application stopped writing to long ago, an encryption key that left along with a former employee. A trial restore swaps guesswork for a measured duration.
The useful question is not whether the jobs are running but when you last restored something and how long it took. We suggest a straightforward exercise: the newest copy goes into a walled-off environment with a stopwatch running. Either you end up reassured, or the broken link identifies itself, and each of those two outcomes pays for the exercise.
It follows from your RTO and RPO. Being usable again by the end of the working day is cheap. Being usable again sixty minutes later requires a warm standby at a hosting provider, which pushes the budget up. We therefore cost the acceptable outage of each application first, and more often than not half the list proves happy with modest protection. A one-off trial restore is billed at the hourly rate of EUR 95/hour excl. VAT; under a continuing agreement it is part of the Start, Business and Premium plans.
An Apply engineer connected remotely, within the response window of your plan: 15 minutes on Premium, 1 hour on Business, 4 hours on Start. The 24/7 IT support service extends cover to nights and weekends. Where a physical action cannot be avoided, replacing a power supply for instance, we talk one of your colleagues or your local contractor through it. Wording stays plain enough for a colleague with no IT background to work through, and a printed copy stays with you.
That is exactly what the trial restores are for: at the rhythm set in the contract, selected data reappears in a walled-off environment and the duration is logged. Once a year, gather the directors and the heads of department around a table and run through a scenario such as “Monday, 7am, the server is displaying a ransom demand”. The gap between the written plan and real instincts shows up within twenty minutes.
There may well be a connection. NIS2 and its transposition into French law deal among other things with business continuity, backups and incident handling, and ANSSI publishes practical guides on all three. Whether your company falls within scope is a matter for your legal adviser rather than for us. If it does, the continuity plan, the log of trial restores and your record of the CERT-FR alerts you act on become usable evidence.
We go through the way your backups are configured, restore one of them for real and measure how long each application genuinely needs to come back.
Message received
An answer follows inside one working day. Report an outage that is stopping people working and it moves ahead of everything else.
Nothing here under that name. Check the spelling, or simply choose the nearest large city instead. Since every engagement runs remotely, whichever you pick changes nothing about what we do for you.
