Legal practice runs on dates. A filing deadline, a hearing listed for ten in the morning, a limitation period that quietly expires at midnight: none of these care whether the office server is working. That makes recovery planning a sensible topic for any practice, and that is what this case study explores. The team below is fictional, assembled to show how the ideas of DisasterRecoveryOps, as presented by the Ops Atlas, apply to a small set of chambers. Read it as an operational illustration, not as legal advice.
The setting
Picture a chambers with eight advocates, three clerks and one part-time IT contractor. Case files are scanned and stored on a shared server. The diary, with every hearing and filing date, lives in a calendar application hosted by a cloud provider. Email runs through a third service. Everything is backed up nightly to a drive in the same office.
The head of chambers believes the arrangement is safe because “everything is backed up”. The contractor is less sure, and suggests a rehearsal. This is the story of what the rehearsal revealed.
Step one: agreeing what the clock measures
The first meeting did not touch technology. The group asked how long the practice could function without its diary and its files before real harm occurred. Hearings that day cannot be postponed on the strength of a dead server. After debate they chose a recovery time objective of two hours for the diary and the active case files, and a one-day target for the archive of closed matters.
They also set a data-freshness limit. For the diary, no more than fifteen minutes of entries could go missing; for files, one working day. In the Atlas this kind of target is a moment on the timeline that data must be recoverable to, and the chambers phrased theirs accordingly: “the state of the diary as it stood at most fifteen minutes before the failure.”
Step two: the rehearsal
On a quiet Saturday the contractor simulated the loss of the office server and the primary diary. A clerk played the part of the first person to notice. Timestamps were recorded for noticing, deciding and restoring.
- Noticing took only a few minutes, because the clerk had been told a failure would happen.
- Deciding took far longer than expected. The head of chambers was the only person who could authorise paying for a temporary cloud server, and was unreachable for a stretch of the simulation.
- Restoring from the office backup drive worked, but the largest folders took much longer to copy than anyone had assumed.
The total exceeded two hours. The pattern matches the sample drill reported in the Atlas, in which the people part of the clock took thirty-one minutes against thirty-three for systems, enough to miss a sixty-minute objective. Our fictional chambers missed by far more, and almost all of the excess was waiting.
Step three: a deadline for the decision
The team adopted a calculation suggested by the Atlas: take the objective, subtract the slowest recent technical restore, and subtract a margin. With a measured restore of ninety minutes and a ten-minute margin, the latest start for a two-hour objective was twenty minutes after the failure. They also authorised the senior clerk to approve the temporary server up to a set limit, with the head of chambers informed afterwards.
Step four: copies that survive the building
The nightly drive sat beside the server. The contractor pointed to a data-centre blaze in 2021, described in the Atlas, after which certain customers found their backups had been stored in the building that burned, and argued for a copy held elsewhere. The chambers added an off-site copy, and kept several earlier versions so that a corrupted file would not simply overwrite its only good twin.
Step five: access that does not depend on the office
During the drill the contractor also discovered that the only administrator password lived in a document on the failed server. The Atlas stresses that recovery credentials belong somewhere outside the failure domain, and recalls a lengthy 2025 cloud disruption in which even the login services were affected. The chambers moved their emergency credentials into a sealed record held by two senior members.
Step six: coming back
The last gap was the return. If the temporary server accepted new documents for a day, how would they be merged into the restored office system? The Atlas calls failback a second migration, and the team wrote a short procedure: preserve the temporary copy, compare against a restored copy rather than the live one, and add the missing files in a controlled order.
Where the clerks fit
The clerks proved decisive. They knew which hearings were imminent and which matters could wait, so they ranked files into tiers: today’s and tomorrow’s hearings first, matters with filings due that week second, closed matters last. This echoes the Atlas sample matrix, which pairs the tightest objectives with the strongest standby arrangements and allows plain backups for the rest.
Clerks also keep public-facing information flowing. Colleagues who rely on our walkthrough of how a litigant can find a case listing online know that court listings give an independent record of hearing dates, so the plan lists them as one way to rebuild the diary. Advocates who have to reach another city quickly if a remote hearing falls through might also bookmark advice on choosing the right rental vehicle for a trip, a small reminder that contingency thinking extends to the journey, not just the server.
One more observation concerns confidentiality. A temporary server that holds client files must meet the same access rules as the original, which means the recovery steps should say who may log in and how access will be removed afterwards. The Atlas does not prescribe legal controls, so the chambers took this point to their own professional-conduct advisers and recorded the answer in the plan.
What the case suggests
None of the chambers’ fixes was expensive. They were conversations, a stopwatch and a few hours of effort. The largest improvement came from naming who decides and by when. The wider discussion, including the Atlas maturity steps from a signed-off objective to routine production switchovers, is on the Ops Atlas entry for DisasterRecoveryOps. A small practice will stop well short of the top, and a documented plan that has been exercised and improved is already a strong position.