Setting Impact Tolerances
A plain account of setting impact tolerances for one important service, what a limit beyond intolerable harm means, and how a firm tests whether the limit holds when disruption strikes.
Line 04: Impact tolerances
Setting a tolerance for each important service: how much disruption, for how long, and with what data loss, before harm becomes intolerable.
The line guide
Working guides on this line
Revising a Tolerance After an Incident
How an incident should lead to revising a tolerance, what counts as evidence that a limit was wrong, and how revision is governed without turning a breach into a quiet rewrite.
Tolerance in Data Loss
What a tolerance in data loss means for one service, how lost, corrupted, or stale data harms users after return, and why recovery of systems is not recovery of data.
Tolerance in Duration
How a tolerance in duration is framed for one service, why length of interruption is only one part of harm, and how timing and user need change what a long stop means.
Tolerance in Volume and Scope
How breadth of users and scale of activity shape a tolerance, why a narrow outage can still cause intolerable harm, and how scope is judged alongside duration.
Tolerances and Customer Harm
How customer harm gives an impact tolerance its meaning, what kinds of customer detriment count, and why harm to customers cannot be reduced to system availability.
Tolerances and Financial Stability Harm
When disruption to one service can harm the wider financial system, how that wider harm differs from customer harm, and why a tolerance must look beyond one firm.
Tolerances and Service Level Agreements
How an impact tolerance differs from a supplier service level agreement, why meeting an SLA does not mean staying inside a tolerance, and how the two are aligned without confusion.
Tolerances for Partial Outages
Why partial outages need their own tolerance thinking, how degraded, inaccurate, or uneven service causes harm, and how a firm judges a service that has not fully stopped.