Convert data storage

About the units

• Bits (bit) – A bit is a binary digit, 0 or 1, and eight bits make a byte. The unit belongs to IEC 80000-13, not to the EU units directive. Bits count in transmission and encryption: broadband is sold in bits per second, and the AES encryption standard works on blocks of 128 bits with keys of 128, 192 or 256 bits. The trap is the symbol. The unambiguous form is “bit”, as in kbit and Mbit, while “b” is easily confused with “B” for byte, a factor of eight apart.
• Bytes (B) – A byte is eight bits and the unit every file size, memory and storage figure is counted in. The quantity is defined in the international standard IEC 80000-13; data volumes are not in the EU units directive and are not regulated as legal units of measurement. One byte holds one character in ASCII, while the Swedish å, ä and ö take two bytes each in UTF-8. The trap is the capital: B means byte and b means bit, a factor of eight apart.
• Kilobytes (kB) – A kilobyte is 1,000 bytes, since the SI prefix kilo means 10³. The binary counterpart is the kibibyte of 1,024 bytes, 2.4% more. The binary prefixes kibi, mebi and gibi were adopted by the International Electrotechnical Commission in 1999 precisely because the kilobyte was being used in two senses. The unit suits small files, text documents and network packets. Write kB with a lower-case k, as with every SI prefix. “KB” with a capital K usually means 1,024 bytes in Windows.
• Megabytes (MB) – A megabyte is a million bytes under the SI prefix mega. The binary counterpart, the mebibyte, is 1,048,576 bytes, 4.9% more, and the megabyte is the very example the American standards body NIST uses: memory makers mean 2²⁰ bytes, storage makers 10⁶. The unit measures photographs, apps, email attachments and mobile data allowances. Two traps. Megabyte against megabit is a factor of eight, so 100 megabits gives 12.5 megabytes per second. And operating systems often write MB while counting mebibytes.
• Gigabytes (GB) – A gigabyte is a thousand million bytes under the SI prefix giga. The binary counterpart, the gibibyte, is 1,073,741,824 bytes, 7.4% more. The unit measures storage, memory and mobile data, and it is not regulated in the EU. Here is the classic trap of the category: drive makers count decimally, so a 1 TB drive is 10¹² bytes, while Windows shows the same drive in binary but writes “GB” – and the figure comes out as 931. Gigabyte against gigabit is a factor of eight.
• Terabytes (TB) – A terabyte is a million million bytes, 10¹², under the SI prefix tera. The binary counterpart, the tebibyte, is 1,099,511,627,776 bytes, 10.0% more – the widest gap among the common prefixes, close to a hundred gigabytes in every terabyte. The unit measures hard drives, SSDs, cloud storage and backups. That is why a drive sold as 1 TB shows up as 931.32 GiB in the file manager while holding exactly what the box promises. TB against Tb, terabit, is again a factor of eight.
• Kibibytes (KiB) – A kibibyte is 2¹⁰ bytes, 1,024 bytes, 2.4% more than a kilobyte. The prefix kibi, symbol Ki with a capital K, was approved by the International Electrotechnical Commission in December 1998 and now sits in IEC 80000-13, and the SI Brochure states that SI prefixes should not be used for powers of two. Yet many programs write only K: in GNU coreutils K means 1,024 but kB 1,000. The old “1.44 MB” floppy disk held 1,440 kibibytes, because its megabyte was 1,024,000 bytes – a blend of both systems.
• Mebibytes (MiB) – A mebibyte is 2²⁰ bytes, 1,048,576 bytes, 4.9% more than a megabyte. According to NIST, memory makers mean 2²⁰ bytes by a megabyte and storage makers a million – the muddle the binary prefixes were meant to clear up. The mebibyte suits sizes that really are powers of two, such as working memory and server memory limits. Kubernetes reads “128Mi” as 128 mebibytes and “128M” as 128 million bytes, and its documentation warns of the trap: “400m” with a lower-case m means 0.4 bytes, not 400 mebibytes.
• Gibibytes (GiB) – A gibibyte is 2³⁰ bytes, 1,073,741,824 bytes. The prefix gibi was adopted by the International Electrotechnical Commission in January 1999 and now sits in the standard IEC 80000-13. The unit belongs where sizes really are powers of two: working memory, virtual machines and Unix tools. A drive sold as 1 TB holds 931.32 gibibytes, and that is the figure Windows shows. The gap against the decimal prefixes widens as you climb: 2.4% for kibi, 4.9% for mebi, 7.4% for gibi and 10.0% for tebi.
• Tebibytes (TiB) – A tebibyte is 2⁴⁰ bytes, 1,099,511,627,776 bytes, 10.0% more than a terabyte – the largest gap among the common prefixes. A drive sold as 4 TB therefore holds 3.64 tebibytes, with nothing missing. The unit is used by system tools and cloud services: Amazon’s block storage gives volume sizes from 1 gibibyte up to 64 tebibytes. The trap is that the same drive shows different figures in different programs. In GNU coreutils T means tebibyte but TB terabyte, and the only difference in the symbol is a B.
• Kilobits (kbit) – A kilobit is 1,000 bits, 125 bytes. The SI Brochure uses the kilobit as its example: SI prefixes stand only for powers of ten, so a kilobit is 1,000 bits and not 1,024. The binary counterpart is the kibibit, Kibit. As a quantity the kilobit survives on memory chips, often in binary. Microchip’s 24LC256 memory chip is called 256 Kbit and has the addresses 0000–7FFF of 8 bits each, 262,144 bits in all. The trap is kb against kB, bit against byte.
• Megabits (Mbit) – A megabit is a million bits, 125,000 bytes. The unit belongs to networks: broadband and mobile data are sold in megabits per second, and the EU open internet regulation requires the contract to explain the minimum, normally available, maximum and advertised speed – but names no unit, so the megabit is trade practice rather than law. The trap is that subscriptions are counted in bits and files in bytes: 100 megabits per second gives 12.5 megabytes per second.
• Gigabits (Gbit) – A gigabit is a thousand million bits, 125 megabytes. As a quantity it appears on memory chips, and there it often means powers of two: Micron’s DDR4 chip marked 16Gb addresses 2³⁴ bits, just under 17.2 thousand million, that is 16 gibibits. Telecommunications, by contrast, counts its prefixes decimally. The Swedish broadband strategy of 2016 aimed for 98% of households and businesses to have access to 1 gigabit per second by 2025. The trap is Gb against GB, gigabit against gigabyte – a factor of eight.