A DisasterRecoveryOps Audit Checklist for Safety Offices

Anyone who has worked inside a safety management system knows the particular quiet of a good audit day. Nobody is panicking. A person with a clipboard asks calm questions, and the answers reveal whether the paperwork matches the practice. I have sat on both sides of that table, and the most useful audits were never the ones that found nothing; they were the ones that found the gap before an event did. This checklist applies the same spirit to a different risk: what happens to your safety records, reporting tools and scheduling systems if the computers behind them stop. The discipline that covers this is DisasterRecoveryOps, and the audit below borrows from it.

Why a safety team has a stake in DisasterRecoveryOps

A safety management system lives on data: hazard reports, investigation files, risk registers, training records, audit findings. If the reporting portal is unreachable on the day someone needs to file an urgent report, the system’s central promise, that concerns get captured, quietly breaks. The Ops Atlas, in its guide to DisasterRecoveryOps, describes the discipline as a continuous operating practice built around recovery objectives, a chosen strategy, a deliberate decision to switch, and regular rehearsal. Those four pieces make a handy frame for an audit.

Before you start

Pick one system to audit, ideally the one whose absence would hurt most on a normal Tuesday. For many safety offices that is the occurrence-reporting tool. Gather three people: someone who uses it daily, someone who administers it, and someone with authority to spend money or change priorities. Block ninety minutes. Bring a stopwatch, because several questions below ask for times you will need to measure, not guess.

Section A: Objectives

  1. Is there a written recovery time target? It should state how long the system may be unavailable before the consequences become unacceptable. If the answer is “as quickly as possible”, the answer is no.
  2. Is there a written data-loss target? The Atlas, echoing the standards body NIST, treats this as a moment on the timeline to which you can wind the system back. Ask: if we restore tonight, what is the most recent hour we could not recover?
  3. Was the target agreed by the people who own the consequences? A target set by the technical team alone often turns out too generous or too strict.
  4. Is the system placed in a tier? The entry argues against demanding zero downtime for everything, and proposes tiers. Record yours, and write down what you are knowingly accepting for lower tiers.

Section B: Decisions

  1. Who may declare that the recovery plan is in effect? Name a person and a deputy. Check both can be reached out of hours.
  2. What is the decision deadline? The Atlas offers a method: take the recovery target, subtract the time of the slowest recent rehearsal, then subtract a margin. The worked sample in the entry arrives at 22 minutes from a 60-minute target, a 33-minute drill and a five-minute cushion. Calculate yours. If the figure is negative, the plan cannot meet its own target even with instant decisions.
  3. Does the plan count the people minutes? The entry stresses that detection and deliberation belong on the clock. Time the last real incident or rehearsal from first symptom to decision, and write the number here: ______.

Section C: Strategy and sequence

  1. Which of the four standard approaches applies? Restore from copies, a minimal always-on core, a scaled-down running copy, or two live sites. Circle one and note why it fits the tier.
  2. Is the sequence written as numbered steps? The Atlas gives the order for moving to a standby: stop the old side accepting changes, promote the data, then redirect users. Check your document follows an order and that every step has an owner.
  3. What proves that the old side has really stopped? Fencing is leakier than it looks. Background jobs and cached connections can bypass a simple switch. Write down how you would verify it.
  4. Is there a plan for the return? Returning to the original arrangement is a second move. If the plan stops at “we are on the standby”, mark it as a finding.

Section D: Access and dependencies

  1. Can you sign in to the recovery tools if the main system is down? The entry warns that the login path itself can share the failure, and advises keeping emergency credentials outside the failure domain. Try it, don’t assume it.
  2. Do you know every store the service needs? The Atlas notes that the real data-loss exposure equals the worst copy delay of any store involved. List them, with the last copy time of each.
  3. Have images, quotas and permissions drifted? A standby that was right a year ago may be missing settings today. Check dates.

Section E: Rehearsal and evidence

  1. When was the last timed drill? If the answer is “never”, you have found the biggest gap.
  2. Did the drill finish inside the target? The entry’s maturity ladder treats two consecutive drills inside the target as the point where a plan counts as exercised.
  3. Were findings tracked to closure? A list of lessons with no dates is a record of intentions.

Scoring what you found

Count your “no” answers by section. The Atlas describes five maturity levels, from objectives that are merely written down, to a plan that passes consistency checks, to repeated drills inside the target, to engineered recovery with a proven return and no improvised steps, and finally to routine production switchovers. Most small safety offices will honestly land on the first or second rung, which is a respectable starting point.

Choose three findings only. The rest can wait. An audit that produces forty actions produces none, because nobody owns forty things. For each of the three, write an owner, a date and the evidence that will show it closed.

The follow-up that makes it real

The entry suggests a quarter-long rhythm that suits a small team, split into three month-long blocks. Block one is about knowing what you have: list the systems, rank them by impact, name who decides and work out the deadline. Block two turns that knowledge into paper and practice, with short written policies, a check for drift between live and standby, and one trial switch away from production. Block three puts the highest-ranked system through a real drill and includes the journey back. Enter all three blocks in the diary before anyone leaves the room.

Examples of findings worth recording

To help you phrase what you discover, here are three typical findings, written in the neutral tone an audit report should have. First: “The recovery document describes technical steps but names no person authorised to start them.” Second: “The only copy of the emergency credentials is stored in the tool whose failure it is meant to address.” Third: “The last drill was timed for the system steps only; the minutes taken to detect and decide were not measured.”

Each is easy to fix and hard to see from inside the team. Notice that none of them requires new software. They require a conversation, a decision and a date, which is why a calm audit is such a cost-effective activity.

When the stopwatch comes out at the next drill, someone will almost certainly say it feels slow and artificial. That is the point. The audit’s value lies in discovering, on a calm afternoon, what you would otherwise discover at the worst possible moment.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top