Understanding the mechanics of traditional hard disk drives
A hard drive stores data on one or more rigid disks called platters, coated with a magnetic layer and spinning on a motor. Tiny regions of that coating are magnetized in one direction or the other, and the pattern of those regions encodes the bits. A read/write head floats on a thin cushion of air a few nanometers above each platter surface and reads or changes the magnetization. An actuator arm swings the heads across the platter to reach different tracks.
Because so much of this is mechanical, a drive can fail in several distinct ways. Heads can wear out or stop working. The heads can touch the platter and scrape the magnetic layer, which is what people mean by a head crash. The spindle motor can seize. The circuit board can fail, or the firmware stored on the platters can become unreadable. Each failure leaves different symptoms: clicking, grinding, silence, a drive that spins but never shows up, or one that shows up with the wrong size.
The practical lesson is that a drive making unusual noises is not a software problem. Repeated power-ups can turn a head problem into platter damage, which is much harder to undo. If the data matters, power the drive off and keep it still. If it works normally, copy what you need to a second device while it still does.
Beginner · hardware · storage · mechanical
What is an SSD?
Solid State Drives and their advantages over traditional drives
A solid-state drive stores data in NAND flash memory chips and has no moving parts. A controller chip manages those chips, maps the logical addresses your computer uses to physical locations in the flash, and spreads writes around so no single block wears out early. That is what makes SSDs fast, quiet, and resistant to drops.
The same design complicates recovery. Flash cells wear with each write, so the controller moves data around constantly, and the mapping between what the computer sees and where the data really sits lives only in the controller's tables. If the controller fails or its firmware corrupts, the chips may be intact but unreadable in any usable order. Many SSDs also encrypt everything internally, even when you never turned encryption on, so reading the chips directly can produce only noise.
Deletion is different too. When the operating system sends a TRIM command, the drive is told which blocks are no longer in use and may erase them in the background. Once that happens, deleted files are usually gone for good, unlike on a hard drive where they often linger until overwritten.
If an SSD suddenly disappears from your computer or shows a strange capacity, stop using it. Keep it powered off, and don't run recovery software that writes to the same drive.
Beginner · ssd · flash memory · storage
RAID Systems Explained
Different RAID configurations and their benefits
RAID, short for Redundant Array of Independent Disks, combines several drives so the computer sees them as one volume. Depending on the level, it improves speed, protects against drive failure, or both.
RAID 0 stripes data across drives. It is fast and uses all the capacity, but there is no redundancy: if any one drive fails, the whole volume is lost. RAID 1 mirrors the same data on two drives, so either can fail and the other carries on. RAID 5 needs at least three drives and spreads parity information across them, so it survives one failed drive. RAID 6 uses two parity blocks and survives two. RAID 10 mirrors pairs of drives and stripes across the pairs, which gives speed and redundancy at the cost of half the capacity.
The thing most people get wrong is that RAID is not a backup. It protects against a drive dying, not against deleted files, ransomware, a failed controller, fire, or a second drive failing while the first is rebuilding. Drives bought together often wear out at similar times, so a rebuild is a risky moment.
If an array loses more drives than it can tolerate, don't force a rebuild or re-initialize it. Note the drive order and which slot each came from before removing anything.
Intermediate · raid · redundancy · storage arrays
Power Surge Damage
How electrical issues affect storage devices
A power surge sends more voltage through a device than its circuits were built to carry. For a storage device, the damage usually lands on the electronics rather than the data itself. A hard drive's circuit board often has protective components, such as a TVS diode and a fuse, that sacrifice themselves to protect the rest. When they do, the drive may not spin up at all, though the platters and heads are frequently untouched.
Typical signs are a drive that is silent after a storm or a sudden outage, a burnt smell, a visible dark spot on a chip, or a drive that powers on in one computer for a moment and then dies. Surge damage can also reach the motor controller and, in worse cases, the preamplifier inside the drive.
Repair often means replacing the failed components or transferring the drive's unique adaptive data from the old board to a matching donor board. That data is stored in a small chip on the board and is specific to each drive, which is why simply buying the same model's circuit board often doesn't work.
If a drive stopped after a surge, don't try different boards or power supplies. Each attempt risks further damage. Protect equipment with surge protectors or a UPS.
Intermediate · power · electrical · pcb
Network Attached Storage
NAS device recovery and RAID reconstruction
A NAS, or network-attached storage device, is a small computer dedicated to storing files and sharing them over the network. Most hold two or more drives, often set up as RAID, and run a Linux-based operating system with its own volume manager.
When a NAS fails, a few different things may have gone wrong. A single drive failure in a redundant setup is usually handled by replacing the drive and letting the array rebuild. Trouble starts when a second drive fails or errors appear during the rebuild. The NAS itself may also die while the drives are fine, or an update, power cut, or controller problem may leave the volume unmounted.
Recovery usually means working from the drives, not the box. Each drive is copied to an image, then the array is reconstructed from the images by working out the RAID level, stripe size, and drive order. Many NAS brands add their own layers on top, so the specifics differ by manufacturer.
Before anything else, write down which drive sat in which bay and label each one. Don't rebuild, re-initialize, or reset the NAS, and don't plug the drives into a different NAS and accept an offer to set up the volume. Shut it down properly and call in someone with experience in your model.
Advanced · nas · network · raid
Tape Drive Recovery
Recovering data from magnetic tape systems
Magnetic tape stores data in a long strip of plastic coated with a magnetic layer, wound on reels inside a cartridge. Data is written in a line along the tape, so reading a file near the end means winding through everything before it. Tape is cheap for its capacity and still widely used for archives and backups, with formats such as LTO and older DAT, DLT, and 8mm cartridges.
Tape problems come from several directions. Cartridges can be stretched, creased, or damaged by a tape drive that failed while it held them. Older tape can suffer from binder breakdown, where the coating gets sticky or sheds. Drives wear out and become hard to find. Some backups were written by software nobody runs any longer, in formats only that software could read.
Recovery therefore has several parts. The cartridge must be physically sound, or be repaired, perhaps by moving the tape into a new shell. A compatible drive has to be found and kept clean. Then the data has to be decoded, which means identifying the backup software and its format, and sometimes rebuilding a catalog.
If you have old tapes, store them cool, dry, and upright, and don't force a stuck cartridge into a drive. Keep notes on which software made them.
Advanced · tape · backup · legacy
Optical Media Recovery
CD, DVD, and Blu-ray disc recovery methods
CDs, DVDs, and Blu-ray discs store data as a spiral track of tiny pits and flat areas, which a laser reads by how the light reflects. Pressed discs, such as movies, have the pits molded into plastic. Burned discs, which most home and office backups are, use a dye layer that the writing laser darkens. That dye is less stable and fades over time and with exposure to light and heat.
Common problems are scratches on the readable side, deep scratches or cracks that reach the data layer, and fingerprints or dirt. Disc rot is the name for the gradual degradation of the reflective layer or dye, which often starts at the edges and leaves a disc that won't read.
Light scratches and smudges can often be handled by wiping gently from the center outward with a soft, lint-free cloth. Resurfacing can help with scratches on the surface, though it removes plastic and can't fix a damaged data layer. A different drive may succeed where the first failed, because drives vary in how well they read marginal discs.
Don't use circular wiping motions, abrasive cleaners, or toothpaste. Hold discs by the edges and keep them in cases, upright, away from heat.
Intermediate · optical · cd · dvd
Firmware Issues
How firmware corruption affects data access
Firmware is the software that runs the drive itself. In a hard drive, it controls the motor and heads, manages bad sectors, and translates the addresses the computer asks for into physical locations on the platters. Part of it is stored on a chip on the circuit board and part on the platters in a protected area called the system area. SSDs have firmware too, and it manages the flash, the wear leveling, and the mapping tables.
Firmware problems have distinctive signs. A drive may spin up normally but never become ready. It may show up with the wrong capacity or a strange name, or freeze when the computer tries to read. A power loss during a write, a bad firmware update, or a bad sector in the system area can cause it.
Fixing it generally needs vendor-level tools. Technicians connect to the drive's service interface, then read, repair, or rebuild the firmware modules, and then image the data. Each family of drives has its own method, which is why the model and family matter.
Do not install firmware updates on a drive that holds data you can't afford to lose, and never flash a drive's firmware as an attempt to fix a failing drive. A wrong firmware file can make the drive permanently unreadable.
Advanced · firmware · corruption · repair
Western Digital Drive Specifics
Understanding WD drive architecture and common failures
Western Digital makes some of the most common hard drives, from small portable My Passport drives to desktop and NAS models. Their basic design is like other manufacturers, but a few quirks come up often in recovery.
WD drives keep adaptive data and firmware modules in the system area on the platters, plus a small chip on the circuit board. Because the board holds unique calibration for that drive, swapping in another board of the same model number usually doesn't work unless that data is moved across. Some models have a known weakness where the heads park on the platter and get stuck, giving a drive that powers up and then clicks or buzzes. Others show up with a very small capacity or a strange name, which often points to firmware trouble.
Many external WD drives, especially My Passport models, can encrypt the contents with a hardware key tied to the USB bridge board. If the bridge board fails, the data on the drive may be unreadable until the original board or its key is recovered, which makes it worth keeping the original enclosure and cable.
If a WD drive acts up, don't keep trying other computers or cables. Don't remove the drive from its external case unless asked to, and don't throw the case away.
Advanced · western digital · firmware · wd
Seagate Drive Recovery
Seagate-specific recovery challenges and solutions
Seagate drives are very common, from desktop and laptop models to external Backup Plus and Expansion drives. Like other brands, they can suffer from head failure, bad sectors, and motor or board problems, but some family-specific faults appear often in recovery work.
One is the so-called BSY state, where a drive is detected but never reports that it is ready, usually because firmware or a firmware module failed to load. Another is a drive that appears with an unexpectedly small capacity or the wrong name. Seagate drives also use system-area firmware and adaptive data, so the board and the platters are matched, and simply swapping a circuit board does not work without moving the right data.
Details differ across generations, so a technician needs the exact family and firmware before working on it.
Do not keep power-cycling a Seagate drive that is stuck or clicking, since each try can cause more damage. Do not try firmware updates to fix it. Record the model number and firmware revision from the label.
Hard drives and SSDs fail in different ways, so recovery looks different for each.
A hard drive stores data on platters. When it fails, the data is usually still on the platters, even if the heads, motor, or board are broken. A technician can replace the failed part, then read the platters. Deleted files often remain until overwritten. Hard drive recovery is mostly a hardware repair job followed by imaging.
An SSD stores data in flash chips managed by a controller. When the controller fails, the chips may be fine, but the controller held the map that links logical addresses to flash locations, and often encrypts the data. Without a working controller or its firmware, rebuilding that map can be difficult or impossible. Wear leveling moves data around constantly, and TRIM lets the drive erase unused blocks in the background, so deleted files can disappear within moments.
That does not mean every SSD is a lost cause. Many SSD failures are firmware or connector problems that can be repaired. But the outcome depends more on the specific model than it does for hard drives.
Whichever type you have, the advice is the same: stop using the device, don't run repair tools, and keep it powered off.
Intermediate · comparison · hdd · ssd
IoT Device Recovery
Recovering data from Internet of Things devices
IoT stands for internet of things: cameras, thermostats, smart speakers, door locks, trackers, sensors, and the small controllers inside appliances. Most hold very little storage, but what they hold can matter, such as a doorbell camera's clips, a sensor's log, or the settings of a device that took hours to configure. Many keep their data only in the vendor's cloud, so the first question is whether the data ever lived on the device at all.
When it did, it is usually on a flash chip soldered to the board, often with a small Linux-based system running from it. Labs get at it in a few ways. Some boards expose a debug port, such as UART or JTAG, that lets a technician talk to the processor. Others need the flash chip read directly with a programmer, either while still on the board or after lifting it off. The raw dump then has to be decoded, because the file system is often a compact one built for flash and not the NTFS or ext4 a computer would show.
Each product is different, and there is no general tool. Encryption, locked bootloaders, and proprietary formats are common. If the device only needs to work again, a factory reset and re-pairing may do it, but that erases the data you may want. Decide which matters first.
Advanced · iot · embedded · firmware
Network Storage Recovery
Recovering data from network storage devices
A network-attached storage box, or NAS, is a small computer that shares its disks over the network. A SAN presents disks to servers as if they were local drives. Distributed systems spread data across many machines. When one fails, the file you want may be split over several disks, formatted with something your laptop cannot read, and organized by software that only the vendor's firmware understands.
Most NAS units from Synology, QNAP, and others use Linux software RAID with a standard file system such as ext4 or Btrfs on top, sometimes with volume management between them. That means the disks can often be read outside the box, once each is imaged and the array is reassembled in the right order. Dead enclosures with healthy disks are the easy case. Failed disks, a rebuild that started on the wrong drive, or a swapped slot order make it harder.
Do not initialize, reformat, or accept an offer from the NAS to create a new volume. Do not keep forcing a rebuild after errors. Label each disk with its slot number before removing it, and keep all of them, including any you think are fine. Encrypted shared folders need their key or password. Business servers also need the network setup written down, such as protocols, shares, and permissions, so restored data can be put back the way people used it.
Advanced · nas · san · network storage
Legacy System Recovery
Recovering data from obsolete systems
Legacy media is anything the world has stopped building equipment for: floppy disks, Zip and Jaz disks, SCSI drives, magneto-optical discs, tape formats that ended decades ago, old IDE drives, and the computers that ran them. People find these in a closet or an estate and want the files, often documents or images from a business or a person who is gone.
The difficulty is usually three things at once. The hardware may no longer work, since rubber parts perish, capacitors fail, and tape binders can shed. The interface may no longer exist on any current computer. And the data format may belong to software nobody can run. A word processor file from an old machine can be perfectly intact and still unreadable until a converter is found.
Recovery tends to go in stages: read the physical media, make an image, then identify the file system and the file types, then convert to something current. Each stage can be done by a different tool. Old media should not be forced. A tape that sheds oxide, or a disk with visible mold, can be damaged further by trying to play it, and some tapes need careful baking or cleaning before one pass. Keep old media cool, dry, and away from magnets, and do not run a questionable disk in a drive you care about.
Advanced · legacy · obsolete · historical
SCSI Drive Recovery
Recovering data from SCSI drives
SCSI is an old family of storage interfaces used in servers, workstations, and some high-end desktops from the 1980s through the early 2000s. A SCSI bus can hold several devices on one cable, each with an ID number, and the two ends of the chain need terminators or signals get reflected and corrupt. Many SCSI drives spun faster than ordinary desktop drives of their day, which is why they turn up in old servers and in equipment such as medical and industrial machines.
When SCSI drives fail, the problems are the usual ones for hard drives: weak heads, surface damage, firmware trouble. What makes them harder is the surrounding gear. The drives come in different connector types, such as 50-pin, 68-pin, and 80-pin SCA, and different bus widths and voltages, so plugging one into the wrong controller can damage it. Many also sit behind hardware RAID controllers that stored the array layout on the card or the disks.
If you have a SCSI server that has stopped, note the order of the drives, leave them in their bays, and do not mix them with others. Pulling drives and reinserting them in a different order can scramble a RAID set. Keep the controller too, since some arrays cannot be read without matching settings. Power it up as little as possible.
Advanced · scsi · enterprise · parallel
SAS Drive Recovery
Serial Attached SCSI drive recovery
SAS, or Serial Attached SCSI, is the interface used by most enterprise hard drives and many server SSDs. It is the successor to parallel SCSI and a cousin of SATA: a SAS controller can usually run SATA drives, but a SATA controller cannot run SAS drives. The connectors are shaped to make sure of that. SAS drives usually have two ports for redundancy, and are built for constant use in storage arrays.
Most SAS drives sit behind a controller or in a storage enclosure with an expander, which is a switch that lets one controller talk to many drives. The array layout, whether RAID 5, RAID 6, or something vendor-specific, is stored by the controller or the array software. When such a system fails, the problem is as often the array as the drive. Drives may be marked offline or foreign, and a controller may offer to clear the configuration. Clearing it erases the very information needed to put the set back together.
If a server with SAS drives fails, do not initialize or clear a foreign configuration, and do not replace and rebuild more than the controller asks. Write down which bay held which drive, and keep the controller and its cache battery if one is fitted. SAS drives are mechanically the same as other hard drives, so clicking or grinding means the same thing: power it down.
Advanced · sas · enterprise · serial
NVMe Recovery
NVMe SSD recovery challenges
NVMe is a protocol for solid-state drives that talks over PCIe lanes directly to the processor, instead of going through the older SATA path. Drives come as M.2 sticks, as add-in cards, and in U.2 server form. They are fast, and they are also very compact, with the controller, flash chips, and sometimes a memory chip packed onto a small board.
Recovery is harder than for a hard drive in a few ways. Data is spread across flash chips by a controller that keeps a translation table, so the chips read alone mean little without it. Many NVMe drives encrypt everything internally, even when you have not turned on a password, and some laptops tie that encryption to the machine's security chip. TRIM, which the system sends to tell the drive what is free, lets the drive erase deleted data in the background, and software recovery of deleted files seldom works.
If an NVMe drive disappears from the system, first check the basics: reseat it, try another slot or enclosure, and look at whether firmware or heat is involved. A drive that reports a wrong name or zero capacity has a controller problem and should be left alone. Do not keep powering it on, since each boot can let it do more internal housekeeping. Where encryption is on, keep the recovery key.
Advanced · nvme · pcie · high speed
M.2 Drive Recovery
M.2 form factor drive recovery
M.2 is a connector and a size, not a technology. An M.2 stick can carry an SATA or an NVMe SSD, and the same slot shape may accept either, or only one. That matters for recovery, because the interface decides which adapters and tools will work. M.2 drives are common in laptops, thin desktops, and tablets, often soldered flat under a cover.
Their compact size has consequences. There is little room for heat to escape, and a drive that overheats may slow down or drop off the bus until it cools. The boards are thin, and flexing, a stripped screw, or a careless swap can crack a trace or a chip. In some laptops the storage is soldered to the board, so a dead board takes the data with it unless the chips can be read directly. Unlike a hard drive, an M.2 stick cannot simply be put into a new case and expected to work if its controller is damaged.
If an M.2 drive stops showing up, power off and let it cool, then try it once in a known-good enclosure that supports its type. Do not run repair tools on a drive that reports as the wrong size. If the machine used encryption such as BitLocker or Apple's, find the recovery key before removing the drive, since the stick may be tied to the board.
Advanced · m.2 · compact · ultrabook
Magnetic Data Recovery
Understanding magnetic storage principles
Hard drives, tapes, and floppy disks store data as tiny magnetized regions. On a hard drive, the platter surface is coated with a magnetic layer, and the head writes by flipping the direction of magnetism in small areas and reads by sensing those flips. The result is a pattern of changes that the drive's electronics turn into bits.
Magnetic media can fail in several ways. Heads can fail or touch the surface. The coating can scratch or flake. A strong magnet or degausser can scramble the data, which is how some drives are destroyed on purpose. Over many years, the magnetization can weaken. Tape can shed its coating or stick when its binder breaks down.
There is a popular idea that overwritten data can be read back from traces with a special microscope. Magnetic force microscopy is real, and it has been used in research, but for modern hard drives with very dense tracks, nobody has shown a practical recovery of overwritten data, and NIST's sanitization guidance treats a single overwrite as sufficient. For deleted but not overwritten data, ordinary recovery is far more realistic.
What you can do is keep magnets away from drives, handle them gently while powered, and avoid long storage in hot or damp places. If a drive has been near a strong magnet, do not test it repeatedly.
Advanced · magnetic · domains · microscopy
Optical Data Recovery
Recovering data from optical media
CDs, DVDs, and Blu-ray discs hold data as a pattern of tiny pits or marks that a laser reads. Pressed discs, like a store-bought movie, have physical pits. Recordable discs use a dye layer that darkens when a laser writes to it, and rewritable ones change the state of a metal alloy. All of them sit under a plastic layer and a protective coating.
Discs fail by scratching, by warping, and by the slow decay of layers, often called disc rot. Sun exposure and heat age the dye of recordable discs. Moisture and poor storage can lift the reflective layer. A scratch on the readable side can sometimes be polished out, but a scratch on the label side often hits the data layer itself and may be permanent.
Reading a damaged disc takes patience. Different drives read differently, and a disc that fails in one may read in another. Tools like ddrescue and specialized optical imaging software read a disc multiple times and fill in gaps, and error correction built into the format repairs small damage.
Copy anything important off old burned discs now. Store them upright in cases, away from light and heat. M-DISC is a recordable format designed for longer life, but it still needs a drive that reads it and a second copy.
Intermediate · optical · laser · error correction
Solid State Principles
Understanding flash memory technology
A solid-state drive stores data in flash memory cells. Each cell holds electrons trapped in an insulated layer, and the amount of charge represents a bit or several bits. Cell types are named by how many bits they hold: SLC holds one, MLC two, TLC three, and QLC four. More bits per cell means more capacity for the price, but the charge levels sit closer together, so reading is less forgiving and the cell wears out sooner.
Flash cells wear because every erase and write damages the insulating layer a little. The drive's controller spreads writes around so no one area wears first, remaps bad blocks, and uses error correction to repair misread bits. Data also leaks over time. A drive left unpowered, especially a worn one stored warm, can start losing data, so an SSD in a drawer is not an archive.
This is why SSD recovery differs from hard drive recovery. The controller keeps a map between the addresses the computer uses and the places the data physically sits. Lose that map and the chips are a pile of scrambled blocks. Encryption in the controller adds another lock.
For you, the practical points are backups and attention. Do not rely on a single SSD for anything important, and replace one that reports errors or health warnings.
Advanced · flash · floating gates · cell types
Read/Write Heads and Head Crashes
The most delicate part of a hard drive
A hard drive's heads sit at the ends of thin arms and float on a cushion of air, or helium, a few nanometers above the spinning platters, closer than any dust particle. Each platter surface has its own head, so a drive with several platters has several heads.
Heads wear and fail with age, drops, and contamination. A weak head may read poorly, causing slowdowns and bad sectors, or not at all, causing clicking. A head crash happens when a head touches the platter, which can scrape off the magnetic coating and ruin the head.
Because the heads are mounted together, a failed head means replacing the whole set with matching parts in a cleanroom. Areas of platter scraped by a crashed head lose their data permanently, which is why shutting down early matters.
When you delete a file on an SSD, the operating system sends a TRIM command telling the drive which blocks no longer hold useful data. The SSD can then erase those blocks in the background during garbage collection, which keeps writes fast.
For recovery, this changes everything. On a hard drive, deleted data stays until it's overwritten. On an SSD with TRIM active, the drive may erase it within minutes, and many SSDs return zeros for trimmed blocks immediately, even before they're physically erased.
TRIM isn't always active: some external enclosures, older systems, RAID setups, and certain file systems don't pass it through. That's why SSD deletions are sometimes recoverable and often not. Either way, stop using the drive immediately.
Shingled and conventional recording, and why it matters
Conventional magnetic recording, CMR, writes tracks side by side. Shingled magnetic recording, SMR, overlaps them like roof shingles to fit more data on each platter. Because writing one shingled track disturbs the next, drive-managed SMR drives use a cache area and an internal translation layer, much like an SSD.
For everyday storage, SMR drives work fine. Under long, heavy writes, such as a RAID rebuild, they can slow dramatically, which has caused problems in NAS and RAID setups.
In recovery, the translation layer adds complexity: if it's corrupted, data has to be located and reassembled before it can be read in order. Check whether a drive is SMR or CMR before buying it for a NAS.
Intermediate · smr · cmr · shingled recording · nas
Helium-Filled Hard Drives
Sealed high-capacity drives
Many high-capacity hard drives are filled with helium instead of air. Helium's lower density lets the platters spin with less turbulence and power, so more platters fit in the same case. The case is hermetically sealed rather than relying on a breather filter.
Helium drives fail in many of the same ways as air-filled drives: heads, firmware, motors, and boards. Opening one in a cleanroom is still possible, but the work has to account for the helium escaping and the drive running in air afterward.
They're common in NAS units and servers, so they turn up often in RAID recoveries. Never pierce or open the seal; there's no home fix.
An external hard drive is a standard internal drive, a small adapter board that converts USB to the drive's own interface, a power supply for desktop models, and a case. Portable drives sometimes build the USB port into the drive's own board instead.
That design explains many failures: a drive that only works with a certain cable, an enclosure that fails while the drive inside is fine, or a drive that's fine until it's dropped. It also explains why removing the drive from its case can backfire on models that encrypt in the adapter.
If an external drive fails, note the model and any noises, keep all the parts together, and don't reformat it when the computer asks.
Many thin laptops, including current MacBooks and some Windows ultrabooks and Chromebooks, have their storage soldered directly to the motherboard instead of on a removable drive. It saves space and power, but the storage can't simply be removed and read in another computer.
If the motherboard fails, the data is trapped on it until the board is repaired. If the storage is also encrypted with keys tied to the board's security chip, as on recent Macs, repair is the only route.
It's worth knowing which kind of laptop you have. With soldered storage, regular backups aren't optional.
RAID 0 splits data into stripes and spreads them across two or more drives, so reads and writes happen in parallel. There's no redundancy: every file is spread across every drive, so if one drive fails, the whole array becomes unreadable.
Recovery means getting a complete image of every member, including the failed one, then working out the stripe size, drive order, and starting offset to put the stripes back together. If a drive is physically damaged, it has to be repaired and imaged first.
RAID 0 is common in two-drive external enclosures set up for speed. Don't reinitialize the enclosure or change the drive order; keep the drives labeled and together.
RAID 1 writes identical data to two or more drives. If one drive fails, the other keeps working, and replacing the failed drive rebuilds the mirror.
The limits: deletions, overwrites, corruption, and ransomware are mirrored instantly to every drive. And if the surviving drive has unnoticed bad sectors, the rebuild can fail or copy errors onto the new drive.
If a mirror member fails, back up the data before rebuilding onto a new drive. If both drives have problems, stop; with two partly failed mirrors, the readable parts of each can often be combined.
JBOD, short for just a bunch of disks, usually means drives joined end to end into one large volume, filling one drive before moving to the next. Some NAS units and enclosures call this spanning or linear mode.
There's no redundancy. If one drive fails, the files stored on it are lost, and files that crossed the boundary between drives are damaged, though data stored entirely on the healthy drives can often be recovered.
The file system usually spans the whole set, so it may not mount with a member missing. Image every drive and assemble them in the correct order before running any recovery.
Intermediate · jbod · spanning · concatenation · nas
Consumer NAS Devices
Small network storage boxes and how they fail
Home and small-office NAS devices are small Linux computers with two or more drive bays. Under the friendly web interface there are usually several layers: Linux software RAID, sometimes a volume manager, and a file system such as ext4 or Btrfs. Many vendors add their own hybrid RAID schemes to mix drive sizes.
Failures come from drives, the NAS hardware, firmware updates, power loss during writes, and ransomware aimed at internet-exposed NAS units. A NAS that won't boot often has perfectly healthy drives.
When a NAS fails, don't accept prompts to initialize, reset, or reinstall, don't move drives between bays, and don't put them into a new NAS without knowing what it will do to them. Label each drive's bay and have them imaged.
Intermediate · nas · network storage · raid · btrfs
Counterfeit-Capacity Flash Drives and Cards
When a big card is secretly small
Some cheap flash drives and memory cards are programmed to report far more capacity than they physically have. They work normally until the real capacity is used up; after that, new data is either discarded or written over earlier data while the device reports success.
The symptom is files that look fine in the folder but are corrupted, empty, or contain pieces of other files when opened, usually the newer ones. Files written before the real capacity ran out can be recovered; data written past that point usually never existed anywhere.
Buy cards and drives from reputable sellers, and check new ones with a capacity-verification tool that writes and reads back the full size before trusting them with anything important.
LTO is the dominant tape format for business backup and archives. Each generation increases capacity, and drives can read only a limited number of older generations, so archives need a plan to migrate before compatible drives become hard to find.
Tapes are written either by backup software in its own format, which needs that software or careful decoding to read, or with LTFS, which makes a tape behave like a drive with files and folders.
Problems include damaged or stretched tape, broken leader pins, cartridge memory faults, and lost backup catalogs. Store tapes upright in a cool, dry place, and record which software and drive generation wrote them.