Types of disaster recovery testing differ in how much of the real environment each one touches. A tabletop is a discussion round a table, while a full failover moves live work to the recovery site. A disaster recovery test checks that recovery works, but the disaster recovery test types below prove different things.
Last reviewed 2026-11-02 by Manish Sharma.
Written for an IT manager or IT head in India or the GCC, this comparison starts from a plain question. A vendor or auditor may ask which kind of test the team has run. Each kind is defined by what people and systems do, then set side by side. A pairing of test with system impact and four questions on the next test follow. The labels are Signisys’ own, built on NIST SP 800-34 Rev. 1, whose pairing appears below, but they are not a standard. Our article on how to run a recovery test covers the method. Our checklist for scoring a recovery test covers the checks.
Types of Disaster Recovery Testing in Short: Seven Kinds on One Ladder
Signisys’ working labels give seven kinds, from lightest to heaviest: plan review, tabletop, runbook walk-through, partial restore, parallel run, partial failover and full failover. In Signisys’ view, a lighter kind costs little and shows less, while a heavier kind shows more and puts more of the live environment at risk. Each rung touches more of the real environment than the one before.
Other sources use different names and different counts for the same ground. NIST separates a test from an exercise. A test uses quantifiable metrics to validate how a system works in an operational environment. It runs as close to real as possible. An exercise, in NIST’s words, is “a simulation of an emergency designed to validate the viability of one or more aspects of an ISCP”. An ISCP is an information system contingency plan. NIST’s control CP-4 in Appendix E lists checklist, walk-through or tabletop, simulation, parallel and full interrupt as example methods. Because names differ, ask what a test does instead of what it is called.
Key Takeaway A test is named by what people and systems do on the day, not by the label on the booking.
For the wider picture, see the Signisys page on testing disaster recovery plans and readiness and the business continuity and disaster recovery solution.
Disaster Recovery Test Types, Defined by What They Do
The disaster recovery test types are easier to compare when each is defined by what people and systems do on the day. None of the definitions depends on a product.
Kinds That Stay on Paper: Plan Review, Tabletop and Walk-Through
A plan review has people read the written plan, the contact list and the system list. They compare each with what exists now. It touches no system, which makes it the lightest rung. This is a Signisys working description.
A tabletop is discussion only. As NIST describes it, a facilitator presents a scenario and asks participants questions, which starts a discussion of roles, responsibilities, coordination and decision making. A tabletop does not involve deploying equipment or other resources.
Advantages
Limitations
A runbook walk-through has people step through the written recovery steps for one system. They read each step against the screen or document but run none of them. This is a Signisys working description.
Kinds That Touch Systems: Restore, Parallel Run and Failover
A partial restore brings backup data back into an area cut off from production, while production keeps running. People then check the data and the time taken. It moves data only, not a running service. This is a Signisys working description.
A parallel run starts the recovery environment while production keeps running. Live traffic stays where it is, even if the recovery site starts. This is a Signisys working description.
A partial failover moves one running service, or one part of the work, to the recovery site until a set window ends. Unlike a partial restore, it carries live work. This is also a Signisys working description.
A full failover moves all live work to the recovery site, then brings it back. This is a Signisys working description; the NIST pairing below puts a system failover to the alternate location at its high-impact level.
The Word Simulation Names Two Different Tests
Sources use the word “simulation” for two different disaster recovery test types, so a booking that says “simulation” can mean either.
The first meaning is a scenario drill. People respond to staged alerts and talk through the plan while no recovery runs. In this article that is a tabletop.
The second meaning is a restore of backup data into an isolated area, and people check the data. In this article that is a partial restore.
Unlike those two meanings, NIST uses the word in its own way. It defines an exercise as a simulation of an emergency, while a functional exercise runs in a simulated operational environment. Check the definition in each document before you rely on it.
What will be restored, where, and will any system be touched? The answer tells you which kind of test is on offer.
Types of Disaster Recovery Testing Side by Side
One table sets the kinds side by side, and it keeps the disaster recovery test types in ladder order.
| Kind | On the day | Systems | Left unproven by a clean result |
|---|---|---|---|
| Plan review | People compare the plan with the real setup. | Nothing is touched. | Whether any step works when run. |
| Tabletop | A facilitator presents a scenario, and people discuss it. | No equipment is deployed. | Whether a restore works. |
| Runbook walk-through | People read each step against the screen, running none. | Nothing changes. | Whether the steps run as written. |
| Partial restore | People restore backup data, then check it. | Data lands apart from production. | Whether the service comes up. |
| Parallel run | People start the recovery environment while production runs. | The recovery site starts, but live traffic stays put. | Whether live work can switch over. |
| Partial failover | People move one service when the window opens. | That service runs from the recovery site. | Whether the whole estate holds together. |
| Full failover | People move all live work, then bring it back. | All live work runs from the recovery site. | Whether recovery works after the estate changes. |
Apart from the tabletop row, which follows NIST, the rows are Signisys’ working descriptions. Each cell in the last column marks the place where the next rung begins.
NIST’s Pairing of Types of Disaster Recovery Testing With System Impact
For US federal information systems, section 3.5.4 of NIST SP 800-34 matches the kind of exercise to the system’s impact level. NIST defines a functional exercise as personnel performing their duties in a simulated operational environment.
| NIST impact level | Kind of exercise | What NIST says it includes |
|---|---|---|
| Low-impact system | A tabletop exercise. | When it runs, it simulates a disruption and includes all main points of contact. |
| Moderate-impact system | A functional exercise. | Its procedures cover an element of system recovery from backup media, as well as all points of contact. |
| High-impact system | A full-scale functional exercise. | It includes a system failover to the alternate location, as well as a full recovery and reconstitution of the system to a known state. |
NIST SP 800-34 covers US federal information systems. For a firm in India or the GCC it is a yardstick and binds no one.
In section 3.5.4 NIST leaves the frequency organization-defined and gives no number, and this article gives none either.
Signisys’ reading, which is not a NIST statement, maps the working labels onto these levels. Plan review, tabletop and walk-through sit at the low end, while partial restore and parallel run sit near moderate. Partial failover and full failover sit at high. A firm sets the impact level of each system itself, so ask the system owner.
Key Takeaway The heavier the system’s impact level, the heavier the test NIST pairs with it.
Choosing Among the Types of Disaster Recovery Testing: Four Questions
Answer these four questions for one system at a time, before booking any test.
What is the system’s impact? Use the NIST pairing above, from a tabletop for low impact to a failover for high.
Which rung has the team already proven? A failover with no restore behind it makes the first miss a live one. For what makes an untested plan fail, see our article on why recovery plans break.
What does a failed test cost? In Signisys’ view, agree the time slot and the point to roll back to before a full failover, because a failed one lands on live work. For the standby decision, see how high availability differs from recovery.
What would a miss teach, and who fixes it? A rung is worth running when its owner will act on the answer. The record and the fix belong to the method article.
Note each system’s impact level, its last proven rung and its owner on one line. This is a Signisys suggestion.
Disaster recovery rests on the plan, the people, the copies and the site, and the test types differ in how much of that each one proves. The seven kinds form a ladder. Answers can differ per system, so one firm can hold a tabletop for one system and a failover for another.
A Tabletop for the Incident Response Plan and for the Disaster Recovery Plan
Both tabletops share a format, a facilitator and a scenario, but they ask different questions. For the incident side, see how to run a tabletop for the incident response plan.
Signisys’ working split is that an incident response tabletop asks who detects, who decides and who tells whom. A disaster recovery tabletop asks which system comes back first, from which copy and against which target. It also asks who calls the vendor.
A ransomware scenario can serve both plans, and why ransomware recovery fails covers the recovery side of such a scenario. One shared scenario can feed both plans, as long as a named person owns each plan’s actions. Before the session, agree which plan the scenario tests, because the same facts lead to different actions.
For how continuity planning differs from recovery planning, read our comparison of continuity and disaster recovery.
What You Still Decide After Choosing a Test, and When to Ask for Help
No type is the best one, and a test does not prove a target by itself. RTO is the recovery time objective, a limit on downtime. RPO is the recovery point objective, a limit on the age of the restored data. A recovery proves them only when someone measures it against them, as setting and meeting those targets explains.
Agree a test rhythm with the system owner. A major change is a reason to test earlier.
Before relying on a kind of test for an audit, check whether your regulator, a client contract or the auditor names one. This comparison does not draw on ISO 22301.
Some systems do not allow a full test. NIST warns that thorough testing of industrial control systems may not be possible during a single testing event. Ask for help when a system has to stay online. Ask too when the recovery path crosses several suppliers, or when an auditor wants evidence the team has no way to produce.
Signisys provides business continuity and disaster recovery planning, as well as consulting and managed cloud services. This describes a Signisys service and is not neutral advice.
Types of Disaster Recovery Testing: FAQ
What are the main types of disaster recovery testing?
Is a simulation test the same as a tabletop exercise?
Does a system’s impact change the types of disaster recovery testing it needs?
Is a parallel run the same as a failover test?
If a tabletop goes well, does it show that a restore works?
Security Insights, Delivered.
Practical, vendor-neutral analysis of threats, tools, and architecture decisions.