Five levels of test

The progression

Level 5 is the only one that tests failback, the trickiest and least prepared operation. Many organisations know how to fail over and discover, after the incident, that they cannot return without a second planned outage.

What each level does not validate

• Level 1 says nothing about feasibility;
• Level 2 says nothing about the technology;
• Level 3 says nothing about inter-system dependencies;
• Level 4 says nothing about reversibility;
• Level 5 says nothing about behaviour under real stress — that is the crisis exercise's job.

A programme stacking three level-1s and one level-2 a year, presented as "four tests performed", gives false assurance. The dashboard must show the level, not just the count.

Timing

A useful test produces a time-stamped chronology with, for each step: start time, end time, planned estimated duration, and the gap.

The items to measure separately:

1. Detection and qualification;
2. Team assembly and decision;
3. Technical preparation (access, entitlements, connections);
4. The restore or failover itself;
5. Functional verification;
6. Service reopening.

Exploiting the gaps

A test report ends with three lists, not with a general conclusion:

• What went better than expected — to be documented, because it is often due to a person rather than a procedure;
• Duration gaps — every gap above 30% becomes an action;
• Blocking points discovered — treated as priority, with a date.

Year-on-year comparison is what gives the programme its value. A test measuring 4 h 47 then 3 h 38 the following year demonstrates improved capability. Two tests concluding "conformant" demonstrate nothing.