Can a penetration test break production systems?
Yes. NIST's testing guide says any time a tester interacts with a system, unexpected halts and other denial of service conditions can follow. The guide's safeguards: an agreed level of intrusiveness, excluded disruptive tests, logged assessor activity, off-hours testing and duplicates of production where they exist.
In detail
NIST SP 800-115, published September 2008, says testing is more intrusive than examination and can impact systems or networks in the target environment. The level of impact depends on the techniques used, from normal network packets that find open ports to specially crafted packets that test for vulnerabilities.
The guide's answer is to decide the acceptable level of intrusiveness before choosing techniques, and to exclude tests known to create denial of service conditions and other disruptions. The processes it lists for reducing risk are skilled assessors, comprehensive assessment plans, logging of assessor activities, testing off-hours, and running tests on duplicates of production systems such as development systems.
The guide treats an assessment as a project, with a plan covering goals and objectives, scope, requirements, team roles and responsibilities, limitations, assumptions, resources, timeline and deliverables, and gives a rules of engagement template in its Appendix B.
| Item | What is agreed |
|---|---|
| Scope | The assets to be assessed, as gathered in the planning phase |
| Techniques | Which are allowed, given the level of intrusiveness agreed; tests known to create denial of service conditions are excluded |
| Timing | Whether testing runs off-hours, which NIST lists among the processes that reduce risk |
| Logging | Assessor activities are logged, which NIST lists among the processes that reduce risk |
| Environment | Whether tests run on production or on a duplicate of production such as a development system |
| Who knows | Whether the IT staff are told (overt testing), which NIST says lets them give guidance to limit the test's impact |
NIST SP 800-115 says organisations need to determine the level of risk they are willing to accept for each assessment and tailor the approach accordingly. Where only production exists, write that accepted level of risk into the rules of engagement before the first packet is sent, and treat any test outside it as out of scope.
For covert testing, NIST says some organisations designate a trusted third party, so the target organisation does not initiate response measures without first verifying that an attack is indeed underway. NIST also says overt testing carries less risk than covert testing, so where production stability matters, choose overt testing.
Where both internal and external testing are performed, NIST says the external testing usually takes place first, which keeps the assessors from acquiring insider information on the network that an outside attacker would not have.

For an Australian organisation
For an APRA-regulated entity, CPS 234 requires a systematic testing program whose nature and frequency are commensurate with, among other things, the rate at which vulnerabilities and threats change, the criticality and sensitivity of the information asset, and the materiality and frequency of change to information assets (paragraph 27).
The same standard requires testing by appropriately skilled and functionally independent specialists (paragraph 30), escalation of deficiencies that cannot be remediated in a timely manner (paragraph 29) and a review of the program at least annually (paragraph 31).
CPS 234 applies to ADIs, general insurers, life companies, private health insurers and RSE licensees. Outside those sectors, NIST SP 800-115 is guidance that nongovernmental organisations may use on a voluntary basis, so the safeguards are whatever the firm writes into its rules of engagement.
