{"canonical_url":"https://engineer.company/glossary/hardware/","content_markdown":"# Hardware\n\n## What it is\n\nHardware is the part of a computing system made of matter rather than of code:\nthe processor that carries out instructions, the memory that holds the work in\nprogress, the disks or flash that keep data while the power is off, the network\ninterface, the power supply and the board that wires them together. Screens,\nkeyboards, cables, switches and routers count as well. A physical server with no\nvirtualisation layer between it and its operating system is often called _bare\nmetal_.\n\nNIST's glossary defines hardware as the material physical components of a system\nand sends the reader on to software and firmware, because the three are defined\nagainst each other. Software, in the same glossary, is programs and data that\nmay be written or modified while they run. Firmware is programs and data held in\nhardware, usually in read-only or programmable read-only memory, so that they\ncannot be written or modified while they run. Firmware is what starts a machine\nbefore any operating system is loaded, and smaller pieces of it run inside\ndisks, network cards and battery controllers. NIST's guidelines on platform\nfirmware treat the hardware and firmware needed to boot and operate a system as\none thing, which they call the platform.\n\nEach part has a capacity that software can use well or badly but cannot raise.\nThe processor does a fixed amount of work per second and is shared by everything\nrunning. Memory is fast, limited and emptied when the power goes. Storage is\nslower and larger, keeps its contents through a power cut, and wears with use.\nThe network interface moves a fixed number of bits per second. A system is as\nfast as whichever of these runs out first for the work it is given.\n\nMost servers today run on hardware nobody at the company has seen. A virtual\nmachine rented from a cloud provider is a share of a physical machine in the\nprovider's data centre: a hypervisor divides that machine's processors, memory\nand disks between the virtual machines placed on it and keeps each to its share.\nThe customer chooses sizes from a price list, and the provider does the cabling,\nthe spare parts and the replacing.\n\n## Why it matters\n\nHardware sets the ceiling. Software decides how much of a machine's capacity\nturns into useful work, but no setting gives a machine more memory than it has.\nKnowing which part is full, the processor, the memory, the disk or the network,\ntells you what to change, and often that the answer is software.\n\nHardware is the slowest and most expensive thing to change. A program can be\nreinstalled in minutes. A part has to be chosen, ordered, fitted and paid for,\nand a larger rented machine costs more every month for as long as it runs. That\nmakes it the last resort for performance, though it is often the first thing\nproposed.\n\nHardware is where things physically wear out. Fans, batteries, hinges,\nconnectors and storage all have working lives. Everything above them can be\nrebuilt from a backup and a written description; the parts themselves have to be\nreplaced.\n\nAnd hardware holds data after software has let go of it. Deleting a file usually\nremoves the file system's record of where the content is, not the content\nitself. A disk that leaves the building, sold, returned under warranty or thrown\nout, can carry much of what was ever written to it.\n\n## What it looks like in practice\n\nThe portfolio's hardware work starts at a repair bench. The account of\n[diagnosing and repairing laptop hardware](/portfolio/diagnosed-and-repaired-laptop-hardware/)\ncovers screens, hinges, keyboards, trackpads, motherboards and power, along with\nsoftware on Linux, Windows and Mac. It puts the skill in the diagnosis: a laptop\nthat will not power on could be the charger, the board or the battery, and the\nrepair is only as good as the guess about what is actually broken. So the care\nwent into finding the real fault before touching anything.\n\nThis site runs on hardware at the other extreme, a machine it has never touched.\nThe credits page describes\n[the one small machine that does all of it](/credits/): a single-core\nDigitalOcean server with 512 MB of memory and a 10 GB disk, standing in Toronto,\nCanada. The same page says Ansible describes that server as code, so it can be\nrebuilt rather than remembered. Off-host backup automation using restic is implemented\nseparately; that implementation alone is not evidence that backups are active. When\nthe hardware belongs to someone else, those two lines are the repair kit.\n\nAnd the note on\n[optimising a database before sharding it](/notes/optimise-before-you-shard/) is\nlargely about when not to reach for hardware. Spreading data across more servers\nspreads slow statements over more hardware without making one of them cheaper.\nRows queueing for a lock are waiting for each other, not for hardware, and a\nbigger machine does not shorten the wait for one row. When many times more\nsessions are active than there are cores, a connection pool fixes that before\nany hardware does. The note's test for splitting is a machine full of good work,\nnot one full of waiting.\n\n## Where it goes wrong\n\n**Hardware is bought for a software problem.** A slow system looks like a small\none. Before paying for a larger machine, find out which part is full. On Linux,\nthree commands give a first answer: how many processors the system can use, how\nmuch memory is in use and available, and five one-second samples of where\nprocessor time goes.\n\n```sh\nnproc\nfree -h\nvmstat 1 5\n```\n\nIn the `vmstat` output, `us` and `sy` are time spent in programs and in the\nkernel, `id` is idle and `wa` is time waiting for storage. A processor that is\nmostly idle while requests wait means the wait is somewhere else: a lock, a slow\nquery, a remote call or too many connections at once. More hardware leaves all\nof those where they were.\n\n**The wrong part is replaced.** A symptom names a place, not a cause. A machine\nthat will not start can be the power, the board or the battery; a slow disk can\nbe a full one, a failing one or a busy one. Swapping parts until the fault goes\naway costs money and hides which part was at fault, so it returns. Test the\ncheapest explanation first, change one thing at a time, and confirm the fix by\nrepeating whatever showed the fault.\n\n**A virtual processor is taken for a physical core.** Sizing a rented machine by\nits count of virtual CPUs assumes each one is a core. DigitalOcean's\ndocumentation, for one, defines a vCPU as one hyper-thread of a processor core.\nOn its shared-CPU plans that hyper-thread may be shared with other virtual\nmachines on the same physical host, and when they are busy a machine can receive\nfractions of one, while memory, disk and network bandwidth stay dedicated.\nMeasure under the load you expect at peak, and move to dedicated processors when\ntimings vary more than the work can tolerate.\n\n**Data leaves with the disk.** Deleting files or formatting a disk can leave its\ncontents readable to anyone with ordinary recovery tools. NIST's guidelines for\nmedia sanitisation, revised in September 2025, define sanitisation as a process\nthat \"renders access to target data on the media infeasible for a given level of\neffort\", and choose the technique by the sensitivity of the information. Decide\nhow each disk will be sanitised before it is bought, and keep a record when it\ngoes.\n\n**Firmware is left out of the updates.** Operating system updates usually leave\nthe firmware in the board, the disks and the network card as it was. NIST's\nplatform firmware guidelines warn that a successful attack on that firmware\ncould leave a system inoperable, perhaps permanently, and build their advice on\nprotecting it from unauthorised changes, detecting the changes that happen and\nrecovering quickly. Apply vendor firmware updates on a schedule, the way the\nrest of the system is patched.\n\n## Further reading\n\n- [hardware — NIST CSRC Glossary](https://csrc.nist.gov/glossary/term/hardware)\n  — the short definition, with its pointers to software and firmware.\n- [DigitalOcean: Choosing the Right CPU Droplet Plan](https://docs.digitalocean.com/products/droplets/concepts/choosing-a-plan/)\n  — what a virtual CPU is at the provider this site runs on, and when shared\n  processors are enough.\n- [NIST SP 800-88 Rev. 2, Guidelines for Media Sanitization](https://csrc.nist.gov/pubs/sp/800/88/r2/final)\n  — how to put the data on a disk out of reach before the disk leaves.\n- [NIST SP 800-193, Platform Firmware Resiliency Guidelines](https://csrc.nist.gov/pubs/sp/800/193/final)\n  — protecting, detecting and recovering the firmware a machine boots from.\n\n\u003chttps://engineer.company/glossary/hardware/\u003e\n","date_modified":"2026-11-12T05:34:23Z","description":"What hardware is, where firmware sits between it and software, what a rented virtual machine gets from the machine under it, and how hardware goes wrong.","formats":{"gemtext":"https://engineer.company/glossary/hardware/index.gmi","gopher":"https://engineer.company/glossary/hardware/gophermap","html":"https://engineer.company/glossary/hardware/","json":"https://engineer.company/glossary/hardware/index.json","markdown":"https://engineer.company/glossary/hardware/index.md","pdf":"https://engineer.company/glossary/hardware/index.pdf"},"kind":"page","language":"en","title":"What is hardware?","url":"https://engineer.company/glossary/hardware/","version":1}