Audit log retention requirements compared

Most frameworks fix a retention floor (PCI DSS 12 months, CIS 90 days, NERC CIP 90 days, HIPAA six years for documentation), while NIS2, DORA and CMMC leave the period to you. Size storage to the longest applicable clock.

On this page

How long must audit logs be kept under each framework?

If you are sizing SIEM storage for several audits, the short answer is that the longest applicable period sets the storage size. PCI DSS requirement 10.5.1 requires audit log history to be retained for at least 12 months, with at least the most recent three months immediately available for analysis. CIS Safeguard 8.10 requires audit logs across enterprise assets to be kept for at least 90 days, and NERC CIP-007-6 Part 4.3 requires the Part 4.1 logs, where technically feasible, to be retained for at least the last 90 consecutive calendar days, except under CIP Exceptional Circumstances. NIS2, DORA, CMMC and the EU delegated rules on ICT logging leave the retention period to the entity, so you must define it from your own risk assessment and the requirements you are answering to.

The split between hot and archived storage matters as much as the total. PCI DSS is the only framework above that names an immediately available window, so the three-month hot tier is usually the figure that drives search performance and index sizing in the SIEM, while the remaining months can sit in cheaper archive storage if you can still restore and search them on request. Confirm with your assessor whether restored archives meet the immediate-availability test. For CMMC, the Level 2 assessment guide expects you to retain logs long enough to investigate a security event, so your own incident response timelines should set that figure. HIPAA's six-year period applies to Security Rule documentation under 45 CFR 164.316, such as policies and risk analyses, not to every security log, so do not size the whole log estate to six years unless your counsel says otherwise.

Scope determines which logs fall under which period. For PCI DSS, confirm which systems are in the cardholder data environment and which connected systems could affect it, since those define the log sources in scope. For NERC CIP, confirm which BES Cyber Systems, associated Electronic Access Control or Monitoring Systems and Physical Access Control Systems are in scope for each applicable requirement, because the 90-day figure applies to the systems the standard covers. Write a retention matrix that maps each log source to every framework that applies, then size storage to the longest period per source. This example is illustrative: a payment platform logging its CDE firewalls for 12 months plus a utility logging only its medium impact systems for 90 days would need different storage for each log class.

Retention floor, scope and proof for each framework

This table compares the retention figure, whether the period is fixed or set by the entity, the log scope, the immediately available requirement and the proof each framework produces. Read each row as a question to answer for your own environment, since the scope of logs in each regime depends on how your systems are classified.

Framework and controlRetention figureFixed or entity-definedLog scope and immediate accessProof produced and where to confirm
PCI DSS 10.5.1At least 12 monthsFixed minimumAudit logs in scope of the cardholder data environment; the most recent three months must be immediately available for analysisReport on Compliance or SAQ with Attestation of Compliance; confirm CDE boundary with your QSA
CIS Safeguard 8.10At least 90 daysFixed minimumAudit logs across enterprise assets; the Safeguard does not set an immediate-access ruleNo certificate; CIS Controls are not a compliance scheme. Confirm asset scope in your own inventory
NERC CIP-007-6 Part 4.3At least the last 90 consecutive calendar days, where technically feasibleFixed minimum, with a CIP Exceptional Circumstances exceptionPart 4.1 logs for Applicable Systems; scope depends on impact categorization under CIP-002CIP compliance audit or spot check by the Compliance Enforcement Authority; there is no CIP certificate
NIS2 (Directive 2022/2555)Not set by the DirectiveEntity-defined through risk-based measuresScope follows the entity's risk assessment and the national transposition; confirm with your national competent authorityRegulator supervision through inspections, security audits, scans and administrative fines; no certificate
DORA Article 10 and Delegated Regulation 2024/1774 Article 12(2)Not set in the Regulation or the delegated actEntity-defined, based on the ICT risk assessmentThe entity defines which events are logged, how long logs are kept and how they are protected; confirm against Commission Implementing Regulation 2024/2690 and your supervisor's expectationsSupervision by competent authorities; the only authority-issued attestation is for threat-led testing under Article 26(7), no logging certificate
CMMC Level 2 (Assessment Guide)Long enough to investigate a security eventEntity-defined, judged by the assessorOrganization defines which audit logs and events to capture and checks that required events are presentCMMC Level 2 assessment by a C3PAO; certificate from the CMMC program. Confirm the retention period your assessor will accept
HIPAA 45 CFR 164.316(b)(2)(i)6 years for Security Rule documentationFixed minimum for documentationApplies to policies, procedures and required documentation; the rule does not state a log-specific period, so confirm whether log records count as documentation for your entityNo certification; HHS does not recognize private certifications. Proof comes from OCR review

Why the retention clocks differ between frameworks

Some frameworks set a number and some make you defend one. PCI DSS 10.5.1, CIS Safeguard 8.10 and CIP-007-6 Part 4.3 each fix a floor you can measure: 12 months with the latest three immediately available for PCI DSS, 90 days for CIS, and at least the last 90 consecutive calendar days for the CIP logs covered by Part 4.1, where technically feasible. Delegated Regulation (EU) 2024/1774 and DORA take the opposite approach. Article 12(2) of the delegated regulation leaves it to you to decide which events are logged, how long logs are kept and how they are protected, in line with your ICT risk assessment. CMMC's Level 2 assessment guide asks something similar: you define the audit events, check that they are captured, and keep logs long enough to investigate a security event. Under these rules, the retention period is only as defensible as the risk analysis behind it, so the documented rationale is the artifact an assessor will test.

HIPAA's six-year figure is a different kind of retention. Under 45 CFR 164.316(b)(2)(i), the Security Rule requires policies, procedures and required documentation to be kept for six years, which is a records obligation rather than a rule for how long system audit logs must be held. Teams that merge the two often size one SIEM tier for six years of raw events, which is usually more storage than a PCI or CIS commitment requires. Keep the two clocks separate in your retention schedule and confirm with counsel and your assessor which records fall under each.

CIP-015-1 is a future obligation, not a current one. FERC approved it on 26 June 2025 in Order No. 907, and it is subject to enforcement from 1 October 2028, with CIP-015-2 extending monitoring to associated EACMS and PACS with an effective date of 1 October 2029. If you operate BES assets in scope, confirm with your Compliance Enforcement Authority which systems and impact categories apply to you, and plan storage for network monitoring data against those dates rather than against the current 90-day floor.

What pushes the retention period and storage up

Retention clocks rarely stay in one place once several audits apply. The drivers below decide how much log history you hold, how fast it must be searchable and how much protected copy you keep, so size them together rather than audit by audit.

Scope of in-scope systems

Every system that stores, processes or transmits cardholder data, or that could affect the cardholder data environment, generates logs that PCI DSS Requirement 10 reaches. For power sector entities, the applicable systems depend on the CIP impact rating of each BES Cyber System, and CIP-007-6 scopes its logging by those categories. Narrowing scope through network segmentation or impact classification usually shrinks the log volume you must keep. Ask your QSA or the Compliance Enforcement Authority which systems fall in scope before you size storage.

Immediately available window

PCI DSS 10.5.1 requires at least 12 months of audit log history, with at least the most recent three months immediately available for analysis. Keeping the recent window in hot, fast-query storage costs more per gigabyte than archive tiers, so the size of that window drives your search infrastructure. Confirm with your QSA how they will test immediate availability, such as whether a query must return results within a set time.

Backup and protection requirements

Logs must be kept and backed up for the predefined period and protected from unauthorized access or changes. That means copies, integrity controls and access restrictions on each tier, and it doubles or triples the footprint of a single retained copy in many designs. Ask the assessor whether backups must be immutable, where they may be stored, and who may restore them, since these answers set the protected copy count.

ICT risk assessment result

Under DORA and its delegated regulation, the entity defines which events are logged, how long logs are kept and how they are protected, based on its own ICT risk assessment. A higher risk rating can justify longer retention or wider event coverage, so the same platform can produce very different storage needs across two firms. Ask your competent authority or auditor which risk outcomes they expect to see documented before setting the period.

Overlapping audits with different clocks

One log store often serves PCI DSS, NERC CIP, CMMC and HIPAA at once, and each sets its own minimum. The longest applicable clock sets the retention for the shared store, while shorter clocks can be met by deleting earlier. HIPAA documentation retention of six years applies to policies and records rather than raw logs, so keep it separate from log storage plans. Map each audit's clock in one table before you buy capacity.

Number of log sources

Storage grows with the number of devices, applications, identity systems and cloud services that emit audit events, and with the event verbosity each one produces. CMMC Level 2 assessors expect you to define which audit logs and events to capture and to confirm that captured records contain the required events, so an enabled-but-noisy source can still add cost. Inventory sources and event types first, then confirm with your assessor which ones they require.

How to confirm your retention requirement before sizing storage

This sequence turns a list of frameworks into one confirmed retention figure per log source, with separate numbers for immediate access, total retention and protection, so storage is sized once against the strictest applicable rule.

  1. Inventory every framework and system in scope

    Start by listing each audit or regulatory regime that applies to your organization, such as PCI DSS, CIS Controls v8.1, NERC CIP, NIS2, DORA, CMMC or HIPAA, and the business unit or entity each one reaches. For each framework, record which systems fall in scope, for example the cardholder data environment for PCI DSS, enterprise assets for CIS Safeguard 8.10, or BES Cyber Systems and their associated access control and monitoring systems for NERC CIP. Owner: the compliance lead with the security architect. Done when every framework has a named scope statement and a list of in-scope log sources.

  2. Record the official retention figure or entity-defined rule

    For each framework, write down the exact retention wording from the controlling source. Some are fixed: PCI DSS Requirement 10.5.1 sets at least 12 months, CIS Safeguard 8.10 sets at least 90 days, and NERC CIP-007-6 Part 4.3 sets at least the last 90 consecutive calendar days where technically feasible. Others leave the period to you: DORA and NIS2 entity-defined rules, and CMMC, where logs must be kept long enough to investigate a security event. Mark each entry as either fixed or entity-defined. Done when every row shows a source citation and a period or a documented rule.

  3. Assign each log source to its governing framework

    Map each log source to the framework that governs it, because one log may fall under several regimes with different periods. A firewall log in the cardholder data environment may carry the PCI DSS period, while the same device in a NERC CIP medium impact environment may carry a 90-day minimum. Record overlaps explicitly. Confirm with the system owner which logs are generated, where they are stored and which are forwarded to the central platform. Done when each source has one or more named frameworks and a storage location.

  4. Separate the immediate-access window from total retention

    Split the sizing model into two numbers. The first is the period that must be immediately available for analysis, which PCI DSS Requirement 10.5.1 sets at the most recent three months. The second is total retention, which may be longer and can sit in colder, cheaper storage if the platform still meets the access expectations. Confirm with your SIEM or log platform team whether archived data can be restored within the working needs of your investigations. Done when you hold one hot-storage figure and one total-retention figure for each source group.

  5. Check protection, integrity and backup requirements

    Retention is not only a duration. Several frameworks require logs to be kept and backed up for the defined period and protected from unauthorized access or changes. NERC CIP specifies protection and backup of retained logs, and DORA's Delegated Regulation leaves protection measures to your ICT risk assessment. Document who can read, alter or delete logs, how backups are tested, and whether storage must sit in a specific region or enclave. Done when a protection and backup standard is written for each log tier and approved by the control owner.

  6. Collect HIPAA documentation retention separately

    If HIPAA applies, treat Security Rule documentation as a separate retention item. Policies, procedures and related documentation must be kept for six years under 45 CFR 164.316(b)(2)(i). This is not audit log retention, but teams that store compliance evidence in the same platform should size for it separately so the six-year commitment is not hidden inside a shorter log policy. Done when the documentation retention item is recorded with its own owner and storage location.

  7. Confirm the period with your assessor or regulator

    Before buying storage, send the mapped table to your QSA, external auditor, or the regulator or Compliance Enforcement Authority that supervises you, and ask them to confirm the periods, the scope and whether your entity-defined rule is acceptable. For NERC CIP, confirm the applicable systems with your Regional Entity. Ask for written confirmation or the names of the reviewers. Done when the confirmed table is signed off and the storage quote is written against it.

  8. Request a storage quote tied to the confirmed table

    Give your SIEM vendor or implementation partner the confirmed table, including per-source volume estimates, hot and total retention figures, protection requirements and backup expectations. Ask the quote to show one-time setup costs separately from recurring storage, licensing and maintenance, and to state which costs rise if retention periods or source volumes change. Done when the quote lists each assumption and can be re-priced if a framework period changes.

Which retention clock applies to which sector

PCI DSS sets a fixed floor for audit log history. Requirement 10.5 has one sub-requirement, 10.5.1, which requires audit log history to be retained for at least 12 months, with at least the most recent three months immediately available for analysis. The standard applies to entities that store, process or transmit cardholder data or sensitive authentication data, and to entities that could affect the security of the cardholder data environment, including merchants, processors, acquirers, issuers and service providers. The PCI Security Standards Council owns the standard, but the payment brands decide which entities must validate and run any enforcement.

  • Fixed floor: 12 months of history under 10.5.1 is the minimum, not a target for every log source.
  • Hot window: the most recent three months must be immediately available, which usually drives how much data sits in fast storage.
  • Version: PCI DSS v4.0.1 is the only active version, so size storage against its wording.
  • Scope: systems that could affect the cardholder data environment often fall under the same clock, which can widen the log sources you keep.
  • Evidence: the assessor will check retention against the ROC or SAQ, so keep the retention configuration as evidence.

Confirm with your acquirer or payment brand which validation level and questionnaire apply to you, and confirm which of your systems count as in scope.

Questions teams ask about log retention before buying storage or outside help

Does the longest retention period decide how much log storage I need?

Mostly yes, but the period is only one input. Storage scales with daily log volume multiplied by how long each log is kept, so the longest period that applies to any system in scope usually sets the size of the archive. The PCI DSS 10.5.1 requirement sets at least 12 months, with the most recent three months immediately available for analysis, which means a hot tier and a cheaper archive tier. CIS Safeguard 8.10 and NERC CIP-007-6 Part 4.3 both set 90 days as a floor, so they rarely drive size on their own. Ask which systems each period applies to, because a single retention rule for every source can overbuy storage.

Can one log store satisfy several audits at once?

Often yes, if you design to the longest and strictest period and tag each source by the frameworks that apply to it. A single store can hold logs for PCI DSS, CIS Safeguard 8.10, NERC CIP and CMMC, provided the retention rules, access controls and backups are set per source. The harder part is evidence: each auditor will ask for different proof, such as the 12-month history with three months immediately available under PCI DSS, or a defined retention set in your own policy for NIS2 and DORA. Confirm with each assessor how they want to sample and export records before you finalize the design.

How much effort does scoping and retention design take?

The effort depends on how many systems are in scope and how many of them already log consistently. Scoping means listing which assets, cardholder data environment systems, bulk electric system cyber assets or enterprise assets produce logs, then deciding which events are captured. Retention design then sets periods per source, the hot and archive split, backup and protection, and who may access or change logs. The EU framework under Delegated Regulation (EU) 2024/1774 leaves these choices to the entity, so the work is a documented decision rather than a lookup. Ask a provider to show the effort split between discovery, design, configuration and evidence, and how many engineer days each takes, in writing.

What drives the cost of log retention up or down?

The main drivers are daily log volume, the number of in-scope sources, the length of the hot tier, the archive format and how often you search old data. Cost recurs every month because storage, compute for search, and backup grow with retention, while the one-time cost covers design, connector work and tuning. Reducing noise at the source, filtering low-value events and moving older records to a cheaper tier lower the recurring bill. Adding frameworks with longer periods, such as HIPAA documentation retention of six years under 45 CFR 164.316(b)(2)(i), raises the cost of anything kept under that rule, so separate documentation from raw logs where you can.

What should I ask for in a written quote for log retention work?

Ask the provider to state the in-scope systems, the frameworks each source must satisfy, the retention period per source, and the hot and archive durations. Request the deliverables in writing, such as a retention matrix, configuration for each source, backup and protection settings, access controls, and evidence samples. Ask what is one-time (design, onboarding, testing) and what recurs (storage, licensing, monitoring, reviews), and whether the storage platform is included or billed separately. DeployOpen plans and implements self-hosted platforms in your own infrastructure and works from a written, fixed scope, so confirm exactly which systems and periods the scope covers before agreeing to it.

Can our own team set up log retention without outside help?

Yes, if you have engineers who know your log sources, storage platform and the requirement wording for each framework. The work is well defined: inventory sources, write the retention matrix, configure periods and tiers, protect and back up logs, and collect evidence. Teams that struggle usually lack time for the scoping decision or have no one who can read the framework wording against their systems. Outside help makes sense when several audits share one store and the design must be defended to assessors. Confirm your skills against the requirement text and the platform documentation before deciding.

What breaks most often in a log retention setup?

The most common failures are sources that never reach the store, retention rules applied to only some systems, and archives that cannot be restored in time for an investigation. Clock drift, missing fields and logs that a process can overwrite also cause gaps that surface during assessment. Check that each in-scope source appears in the store, that the period is enforced by the platform rather than by memory, and that a sample restore from the archive works. Test the restore before an audit, because an unreadable archive is a finding even when the data exists.

Can one SIEM store satisfy several audits?

Yes, in most cases, but only if you write one retention policy that meets the longest applicable clock and then scope each source to the rules it actually falls under. The longest clock in the topic facts is the PCI DSS requirement 10.5.1 of at least 12 months, with at least the most recent three months immediately available for analysis. A single SIEM tier sized for that rule will usually cover CIS Safeguard 8.10 (90 days) and the 90-day CIP-007-6 Part 4.3 window for systems that fall in scope. The hot-storage sizing comes from the three-month rule, so confirm with your QSA and your SIEM vendor how "immediately available" is defined and whether it requires indexed search rather than archived files.

Keep the scopes separate even when the storage is shared. Tag each log source with the frameworks it serves, such as cardholder data environment systems for PCI DSS, enterprise assets for CIS, or high and medium impact BES Cyber Systems for NERC CIP. Retention rules differ by source: CIP logs are tied to the Applicable Systems tables in CIP-007-6, EU entities must define their own periods under Delegated Regulation (EU) 2024/1774 Article 12(2), and HIPAA's six-year clock applies to documentation rather than log events, so it should not drive SIEM sizing at all. Ask each auditor which sources they treat as in scope before you extend a retention period across the board.

Protection matters as much as duration. Each framework that sets a retention period also expects logs to be protected from unauthorized access or change, so the policy should name who can alter retention settings, how deletions are logged and how backups are verified. Write the policy as a controlled document, approve it, and record each exception with its owner and expiry date. Auditors will ask for the policy, the configured settings and evidence that they match, so generate that evidence from the platform rather than from memory.

How DeployOpen fits into a log retention project

DeployOpen plans, implements, documents and supports self-hosted open-source platforms inside your own cloud accounts, data centers or air-gapped networks. For a log retention project, that means the log pipeline, the storage tier that holds the logs, the access controls on that storage and the backup jobs that protect the retained records can be scoped and built in your environment rather than in a vendor's hosted service. DeployOpen also prepares evidence for compliance assessments, such as configuration exports, retention settings and access records, but it does not issue audit reports, attestations or certifications. Those come from your assessor, certification body or regulator. Most engagements are fixed-scope, so the retention design, deliverables and timelines are agreed in writing before work starts, and any ongoing support is handled as a separate retainer.

The retention periods themselves stay with you. The retention decision belongs to your assessor or regulator, and it differs by framework: PCI DSS sets a fixed minimum, CIS Safeguard 8.10 sets a floor, while NIS2, DORA and the EU delegated rules leave the period to your own risk assessment. Before scoping, ask each assessor or regulator to confirm the period in writing, record which systems and log sources each period applies to, and identify the longest requirement that governs each store. That single list of periods, sources and owners is what turns into the storage sizing and the scope statement in the quote. DeployOpen can then build against that written scope, while the compliance decisions remain documented on your side.

How DeployOpen helps scope and implement log retention

DeployOpen plans, implements and documents self-hosted log storage and retention inside your own cloud accounts, data centers or air-gapped networks. We start from the retention periods each framework in scope sets, and from the retention periods your own ICT risk or security assessment defines where a framework leaves that choice to you.

How we scope

  • Inventory which log sources, systems and assets each in-scope framework covers before choosing any retention period.
  • Map every retention clock, including the 12-month PCI DSS period and the 90-day CIS and NERC CIP periods, to a single storage plan.
  • Agree in writing which periods you define yourself and who in your organization signs off on them.

How we build

  • Configure log collection, retention tiers and immediately searchable storage in the SIEM or log platform you run.
  • Set protection controls so retained logs stay safe from unauthorized access or change, with backups kept to your defined period.
  • Test retrieval of the oldest retained records to confirm they can be searched and used in an investigation.

How we hand over

  • Document the retention matrix by framework, log source and storage tier, including the assumptions behind each period.
  • Provide runbooks for storage expansion, archive restores and retention changes when a framework version or scope changes.
  • Deliver evidence packs that show configured retention and access controls for your assessors to review.

How we support

  • Offer ongoing support, upgrades and reviews of retention settings as frameworks and in-scope systems change, through a retainer.
  • Revisit storage sizing when new log sources are added or when a regulator or assessor asks for a different period.
  • Confirm with your auditor or regulator which retention periods they accept before finalizing the design.

Sources and further reading

Related

Talk to our team.

Tell us what you're working on, whether it's a deployment, an audit, a security test or a cyber range. You'll speak with an engineer who can help you scope it.

  • 30-minute call: free, with no obligation.
  • NDA on request: we can sign before you share details.
  • Clear next steps: a scope and plan after the call.