A forensic image that was taken the wrong way, in the wrong format, without verification or after the encryption key was lost cannot be repaired later, and the questions it should have answered stay open for good. Disk imaging is the process of creating a complete, verifiable copy of a storage device or volume so that analysis happens on the copy and the original stays unchanged. This guide covers when to take a full image instead of a triage collection, how to choose between raw, E01 and other formats, how write blocking works, how to deal with BitLocker and other encrypted volumes, how to acquire cloud disks through snapshots in AWS and Azure, and how to verify and document every image. It is written for incident responders, forensic analysts and the IT and cloud teams who are asked to "grab an image" during an incident. For fast targeted collection, see the KAPE and Triage Collection Guide.
Key takeaways
- Decide image or triage first. Full images preserve everything, including deleted data and unallocated space; triage collections are faster. Many incidents need triage on many hosts and full images of a few.
- Prefer E01 with metadata and hashes for evidential work, and raw when tools or speed demand it. Either way, hash the source and the image and record both.
- Write block every physical acquisition. Use a write blocker tested against the NIST CFTT Hardware Write Block specification.
- Secure the key before you power down. For BitLocker, FileVault or LUKS, retrieve the recovery key or image the live, unlocked volume first; a powered-off encrypted disk without a key is noise.
- In the cloud, the snapshot is the image. Snapshot volumes, copy them to an isolated forensic account with a customer managed key, and analyse from volumes created there.
Image or triage: choosing the acquisition type
| Acquisition type | What it captures | Use when | Trade-off |
|---|---|---|---|
| Physical image | Every sector of the device, including unallocated space and partition tables | Legal or disciplinary cases, suspected wiping or anti-forensics, key servers in a major incident | Slow; large storage needs; not possible for most cloud and SaaS |
| Logical (volume) image | One partition or volume, often decrypted | Encrypted systems imaged live; when only one volume matters | Misses data outside the volume |
| Logical file collection (for example L01 or a zip of selected files) | Specific files and folders with metadata | Mail stores, user folders, file servers | No deleted or unallocated data |
| Triage collection | Targeted artefacts: event logs, registry hives, $MFT, browser data, prefetch | Scoping many hosts quickly | Answers known questions only |
| Cloud snapshot | Block-level copy of a cloud disk at a point in time | IaaS virtual machines in AWS, Azure or Google Cloud | No physical layer; captured while running unless stopped |
| Memory capture | Volatile RAM contents | Live malware, encryption keys in memory, network connections | Must be done before shutdown; see the Memory Forensics Guide |
Order of volatility: capture memory first if the host is running and the case may need it, then the disk. Do not shut down an encrypted laptop before you have a key or a live image. Record why each choice was made, because the reasoning will be questioned later.
Image formats: raw, E01 and alternatives
| Format | Structure | Strengths | Weaknesses |
|---|---|---|---|
| Raw (dd, .img, .001 segments) | Bit-for-bit copy with no container | Universal tool support; mountable directly; simplest to verify | No embedded metadata or hash; full size on disk; segments are only split files |
| EWF E01 (Expert Witness Format) | Segmented container with case metadata, compressed chunks and checksums | Case number, evidence number, examiner and notes in the header; zlib compression; per-chunk Adler-32 checks; stored MD5 or SHA-1 of the acquired data | Proprietary origin, documented through reverse engineering (libewf); slower to write |
| EWF2 Ex01 | Successor introduced with EnCase 7 | Newer features in supporting tools | Less widely supported than E01 |
| L01 or Lx01 | Logical evidence file containing selected files | Keeps file metadata and hashes | Not a disk image |
| AFF4 | Open, container-based format | Open specification; designed for large and sparse images | Uneven tool support |
The libewf documentation describes how E01 stores metadata in header sections, compresses data in chunks (32,768 bytes by default) with zlib, protects sections and uncompressed chunks with Adler-32 checksums, and splits images into segment files named .E01 to .E99 and then .EAA onwards. It also notes that the E01 password field does not protect the data: protect images with storage encryption and access control instead.
Decision guidance: choose E01 for evidence that may be examined by others or presented in legal proceedings, because the metadata and embedded hashes travel with the image. Choose raw when the analysis toolchain needs it, when imaging to fast local storage for immediate analysis, or when converting cloud snapshots. Converting later is possible, but always re-verify the hash after conversion.
Write blocking and physical acquisition
A write blocker sits between the source drive and the acquisition machine and allows read commands while blocking anything that could modify the drive. NIST's Computer Forensics Tool Testing (CFTT) programme publishes a Hardware Write Block specification (version 2.0 is current) with a test plan, and test reports on specific devices hosted by DHS. Check a device's report before you rely on it, and test it yourself after firmware updates.
- Document before touching: photograph the device, labels and serial numbers; record the time and who is present.
- Connect through the write blocker (SATA, SAS, NVMe or USB as appropriate). Software write blocking is acceptable when hardware is not available, but record the method and test it.
- Identify the full device: confirm the reported size matches the label, and check for hidden areas such as a Host Protected Area that may need to be included.
- Hash the source where the tool supports it (SHA-256, with MD5 as a secondary hash if your lab requires it).
- Acquire to clean, verified destination media, using a tool that logs read errors. Record bad sectors rather than retrying indefinitely on a failing drive.
- Verify by hashing the image and comparing it with the acquisition hash; record both values in the acquisition log and the chain of custody record.
- Make a working copy and store the master image read-only.
Solid state drives: SSDs can erase blocks in the background after files are deleted (TRIM and garbage collection) even when a write blocker is connected, because this happens inside the drive. Image SSDs early, and expect less recoverable deleted data than on hard drives. Hash mismatches between two reads of a failing or self-managing drive should be explained in the notes, not hidden.
Encrypted volumes: BitLocker, FileVault and LUKS
Full disk encryption is now the default on most corporate laptops, which changes the acquisition order. Three routes:
- Live logical image: if the machine is running and unlocked, image the decrypted volume live (after memory capture). This is often the only route for systems with keys held only in a TPM.
- Recovery key and physical image: take a physical image of the encrypted disk and decrypt it during analysis with the recovery key. Microsoft documents that BitLocker recovery passwords are 48-digit numbers, stored for Entra joined devices as a
bitlockerRecoveryKeyresource retrievable from the Microsoft Entra or Intune admin centres and Microsoft Graph, and for Active Directory joined devices asms-FVE-RecoveryInformationobjects under the computer object. The key ID shown on the recovery screen identifies which stored password applies. - Key from memory: a memory capture of a running system may contain volume keys that forensic tools can use.
Retrieving recovery keys is a privileged action: restrict it to named forensic and help desk roles, log every retrieval, and record it in the case notes. An attacker with the same rights can read the same keys. See the Disk Encryption Guide: BitLocker and FileVault and the LUKS Disk Encryption Guide.
Cloud disks: snapshots as forensic images
In IaaS, there is no physical disk to image. The snapshot is the acquisition.
| Step | AWS (EBS) | Azure (managed disks) |
|---|---|---|
| Capture | Create an EBS snapshot of each volume attached to the instance. AWS documents snapshots as incremental, stored in S3 you cannot access directly, with each snapshot containing everything needed to restore a new volume | Create a snapshot of each managed disk; Microsoft describes a snapshot as a full, read-only copy of a managed disk |
| Isolate | Share or copy the snapshot to a dedicated forensic account. Encrypted snapshots can only be shared if encrypted with a customer managed KMS key, and that key must also be shared; snapshots encrypted with the default AWS managed key cannot be shared | Copy the snapshot to a dedicated forensic subscription or resource group with restricted access |
| Region | Snapshots are constrained to their Region; copy to another Region before sharing there | Keep in the same region unless policy requires otherwise; record any copy |
| Analyse | Create a volume from the snapshot in the forensic account and attach it to an analysis instance, mounted read-only | Create a managed disk from the snapshot and attach it to an analysis VM as a data disk, mounted read-only |
| Export (optional) | Convert to raw or E01 from the attached volume if offline analysis is needed; hash the result | As AWS |
Practical points: tag snapshots with the case ID; capture the instance metadata, security groups, IAM role and network interfaces at the same time; snapshot all volumes as close together as possible; and remember that a running instance gives a crash-consistent copy at best. The roles that create and share snapshots are powerful non-human and human identities: a snapshot shared to the wrong account gives that account all of the data. See the Cloud Forensics Guide.
Verification, documentation and chain of custody
ISO/IEC 27037:2012 gives guidelines for the identification, collection, acquisition and preservation of digital evidence across storage media, mobile devices, networked systems and similar devices; ISO's page shows it under systematic review since 2023, so check the current edition before citing it in procedures. NIST SP 800-86 covers integrating forensic techniques into incident response from an IT perspective. Both expect that the acquisition can be repeated or explained and that integrity can be shown.
- Acquisition log: case ID, evidence ID, device make, model and serial, capacity, interface, write blocker used, tool and version, start and end time, examiner, hash values, errors.
- Hashes: record the source hash (where computed) and the image verification hash; any mismatch is explained.
- Chain of custody: every transfer of the original device and master image. Use the Chain of Custody Form Template.
- Storage: master images on encrypted, access-controlled storage with logging; working copies only for analysis.
See the Evidence Handling and Chain of Custody Guide and the ISO/IEC 27037 Digital Evidence Guide.
Implementing an acquisition capability
- Prepare kits: tested hardware write blockers for each interface, wiped and verified destination drives, an acquisition laptop with current tools, cables, labels, forms.
- Prepare the cloud: a dedicated forensic account or subscription, a customer managed key for snapshot sharing, a pre-approved role for snapshot creation and sharing, and an analysis image with tools installed. Test the whole path before you need it.
- Prepare key access: confirm BitLocker and FileVault recovery keys are escrowed for every managed device, and who can retrieve them out of hours.
- Write the procedures: one page each for physical, live, encrypted and cloud acquisition, with the decision table above.
- Test: image a test laptop, a test server and a test cloud instance each quarter, verify the hashes, and time each one. Include acquisition tasks in exercises. See the Cyber Incident Exercises Beyond Tabletop Guide.
Small versus large organisations: a small organisation may keep one write blocker kit, rely on cloud snapshots and an incident response retainer for physical imaging, and focus on key escrow and the forensic account. A large organisation needs kits at each major site, trained local staff and an evidence storage service. See the Forensic Readiness Guide.
Detection and measurement
- Time to acquire: from request to verified image, for laptop, server and cloud instance.
- Verification rate: percentage of images with a recorded and matching verification hash (target 100%).
- Key escrow coverage: percentage of encrypted managed devices with a retrievable recovery key.
- Documentation completeness: percentage of acquisitions with full acquisition log and chain of custody.
- Snapshot access monitoring: alerts on snapshot sharing to accounts outside the forensic account list, and on recovery key retrieval outside a case.
Framework mapping
| Element | Reference |
|---|---|
| Integrity and provenance of collected incident data | NIST CSF 2.0 RS.AN-07 |
| Recording investigation actions with integrity | NIST CSF 2.0 RS.AN-06 |
| Root cause analysis using the evidence | NIST CSF 2.0 RS.AN-03 |
| Incident handling including forensic analysis | NIST SP 800-53 Rev. 5 IR-4 Incident Handling |
| Protecting stored evidence and logs | NIST SP 800-53 Rev. 5 AU-9 Protection of Audit Information, SC-28 Protection of Information at Rest |
| Collection of evidence | ISO/IEC 27001:2022 Annex A 5.28; ISO/IEC 27037:2012 |
Common failure patterns
- Powering off encrypted laptops before retrieving a key or capturing memory, leaving an unreadable disk.
- No verification hash, so the integrity of the image cannot be shown later.
- Imaging cloud instances by copying files from inside the compromised VM instead of snapshotting from the control plane.
- Sharing snapshots with broad access or using the compromised account for analysis.
- Untested write blockers or firmware updates applied without retesting.
- Rebuilding before imaging, because restoring service was prioritised and nobody owned evidence collection.
Practitioner checklist
- Use a written decision table for image versus triage versus snapshot.
- Standardise on E01 for evidential images and record when raw is used and why.
- Keep tested hardware write blockers for every interface you support.
- Capture memory before shutdown when encryption or live malware is involved.
- Confirm recovery key escrow for every encrypted managed device and restrict retrieval.
- Build a forensic cloud account with a customer managed key and test snapshot sharing.
- Hash every source and image and record both in the acquisition log.
- Store master images read-only on encrypted, logged storage and analyse only copies.
- Time a test acquisition of each asset type every quarter.
Standards and references
- ISO: ISO/IEC 27037:2012 Guidelines for identification, collection, acquisition and preservation of digital evidence
- NIST: SP 800-86 Guide to Integrating Forensic Techniques into Incident Response
- NIST: CFTT Hardware Write Block Specifications and Test Reports
- libyal: Expert Witness Compression Format (EWF) Specification
- Microsoft Learn: BitLocker Recovery Overview
- AWS: Amazon EBS Snapshots
- AWS: Share an Amazon EBS Snapshot
- Microsoft Learn: Create a Snapshot of a Virtual Hard Disk
Related NHI Mgmt Group resources: KAPE and Triage Collection Guide · Digital Forensics (DFIR) Guide · Cloud Forensics Guide · Memory Forensics Guide · Autopsy and The Sleuth Kit Guide