Comparison of archive formats

There are many popular computer data archive formats for creating and maintaining archive files. The tables below compare many popular archive formats.

Features

The table compares various features column-by-column in the table below:

Purpose

Archive formats are used for backups, mobility, and archiving. Many archive formats compress the data to consume less storage space and result in quicker transfer times as the same data is represented by fewer bytes. Another benefit is that files are combined into one archive file which has less overhead for managing or transferring.

There are numerous compression algorithms available to losslessly compress archived data and some algorithms work better (smaller archive or faster compression) with particular data types.

Archive formats are also used by most operating systems to package software for easier distribution and installation than binary executables.

Filename extension

The DOS and Windows operating systems required filenames to include a three-character extension to identify the file type and use. Filename extensions must be unique for each type of file. Many operating systems identify a file's type from its contents without the need for an extension in its name. However, the use of three-character extensions has been embraced as a useful and efficient shorthand for identifying file types—both for computer software, and for humans.

Integrity check

Archive files are often stored on magnetic media, which is subject to data storage errors. Early tape media had a higher rate of errors than is expected for magnetic media today. Many archive formats contain extra data embedded in the files in order to detect data storage or transmission errors, and the software used to read the archive files contain logic to detect errors.

Recovery record

Many archive formats contain redundant data embedded in the files in order to detect data storage or transmission errors, and the software used to read the archive files contain logic to detect and correct errors.

Encryption

In order to protect the data being stored or transferred from being read if intercepted, many archive formats include the capability to encrypt the data. There are multiple mathematical algorithms available to encrypt data.

Comparison

Containers and Compression

Format	Filename extension	Created by	Introduced in	Based on	Compression	Integrity check	Recovery record	Encryption supported	Unicode filenames	Modification date resolution	Pre-processing
Archive (ar)	.a	CSRG	?	Original	No	No	No	No	No	1 s	?
cpio	.cpio	Bell Labs	1983 Unix System V	?	No	Partial, select formats only	No	No	No	1 s	?
Shell Archive (shar and makeself)	.shar, .run	?	1994 4.4BSD	Original	No	Yes, commonly MD5	Partial	Partial	Partial	arbitrary (typically 1 s)	?
Tape Archive (tar)	.tar	Bell Labs	1975 Version 6 Unix	?	No	Partial, metadata only. Full integrity providable by filters such as gzip.	No	No	Optional	1 s	No
Extended TAR format (pax)	.tar	OpenGroup	2001	Sun proposal + TAR	No	metadata	No	No	Yes	arbitrary (typically 1 ns)	?
BagIt	-	The Library of Congress	2007	file system	No	Yes	No	No	Yes	No	?
7z	.7z	Igor Pavlov	2000	LZMA	Yes	Yes, CRC32	No	Yes, AES-256	Yes	1 ms (maybe better?)	Yes
ACE	.ace	Marcel Lemke	?	?	Yes	Yes	Yes	Yes, Blowfish	Yes	?	?
AFA	.afa	Vicente Sánchez-Alarcos	2009	Original	Yes	Yes	Yes	Yes, AES and CAST	Yes	?	?
ARC	.arc	Thom Henderson (SEA)	1985	?	Yes	CRC16	No	weak XOR only	No	2s	?
ARJ	.arj	Robert Jung	1991	AR001 and AR002	Yes	Yes	Yes	weak XOR with initial constant	No	?	?
B1	.b1	Catalina Group Ltd	2011	LZMA	Yes	Yes	No	Yes, AES	Yes	?	?
Cabinet	.cab	Microsoft	1992 Windows 3.1	DEFLATE	Yes	Optional PKCS7 Authenticode signature	No	Optional (with SDK)	Yes	2 s	?
Compact File Set	.cfs	Joe Lowe (Pismo Technic Inc.)	2008	ZIP/LZMA	Yes	Yes	?	Yes	Yes	?	?
Compact Pro	.cpt	Bill Goodman	1990 (as "Compactor")	Original	Yes	Yes	No	Yes	?	?	?
Disk Archive (DAR)	.dar	Denis Corbin	2002	Original	Yes	Yes	Yes	Yes	Yes	1 µs	Yes
DGCA	.dgc	Shin-ichi Tsuruta	2001	GCA	Yes	Yes	Yes	Yes	Yes	?	?
FreeArc	.arc	Bulat Ziganshin	2006	LZMA, PPMD, TTA	Yes	Yes	Yes	Yes, AES, Blowfish, Twofish and Serpent	Yes	?	?
LHA (also LZH)	.lzh, .lha	Haruyasu Yoshizaki	1988	Frozen	Yes	Only on recent LHA releases	No	No	No	1–2 s	?
LZX	.lzx	Jonathan Forbes and Tomi Poutanen	1995	LZ77	Yes	Only on recent LZX releases	?	?	?	?	?
Sparc	.arc	David Pilling	1989	?	Yes	?	?	?	?	?	?
WinMount format	.mou	?	2007	?	Yes	Yes	Yes	Yes	Yes	?	?
Macintosh Disk Image	.dmg	Apple Computer	2001 Mac OS X	Original	Yes	Yes	?	Yes	?	?	?
Partition Image (PartImage)	.partimg	François Dupoux and Franck Ladurelle	2000	?	Yes	?	?	?	?	?	?
PAQ Family (Several formats)	.paq#*, .lpaq#*	Matt Mahoney	2002–2006	Original	Yes	?	?	?	?	?	?
PEA	.pea	Giorgio Tani	2006	Original, Deflate based compression	Yes	Yes Adler32, CRC32, CRC64, MD5, SHA1, RIPEMD-160, SHA256, SHA512, Whirlpool	No	Yes Authenticated Encryption, AES128 and AES256 in EAX mode	Yes system dependent	Yes arbitrary	?
PIM	.pim	Ilia Muraviev	2004–2008	Original	Yes	Yes	No	No	Yes	No	?
Quadruple D	.qda	Taku Hayase (aka sandman)	1997	?	Yes	?	?	?	?	?	?
RAR	.rar	Eugene Roshal	1993	Original	Yes	Yes, CRC32, BLAKE2	Yes, Reed-Solomon	Yes, AES-256	Yes, UTF-8	2 s, 1 s, 6.5536 ms, 25.6 µs or 100 ns	Dropped
RK	.rk	M Software, Ltd.	2004	Original	Yes	Yes	No	Yes, AES, Square, Twofish	Yes	1 s	?
StuffIt (also SIT)	.sit	Raymond Lau	1987	?	Yes	?	?	Yes	?	?	?
StuffIt X (also SITx)	.sitx	Aladdin/Allume Systems	2002	?	Yes	?	Optional	Yes, RC4,Blowfish, AES,DES	Yes	?	?
UltraCompressor II	.uc .uc0 .uc2 .ucn .ur2 .ue2	Nico de Vries	1992–1996	LZ77 and Huffman coding	Yes	Yes	Yes	Yes, triple DES	?	?	?
Windows Image	.wim	Microsoft	?	Original	Yes	Optional	?	No	Yes	?	?
ZIP (also PKZIP)	.zip	Phil Katz	1989	DEFLATE	Yes	Yes	No	Yes, AES	Yes	2 s	?
ZPAQ	.zpaq	Matt Mahoney	2009	PAQ	Yes	Yes, SHA-1	No	Yes, AES-256	Yes	?	?

Notes

^1 While the original tar format uses the ASCII character encoding, current implementations use the UTF-8 (Unicode) encoding, which is backwards compatible with ASCII.
^2 Supports the external Parchive program (par2).
^3 From 3.20 release RAR can store modification, creation and last access time with the precision up to 0.0000001 second (= 0.1 µs). [1] [2]
^4 The PAQ family (with its lighter weight derivative LPAQ) went through many revisions, each revision suggested its own extension. For example: ".paq9a".

Software Packaging and Distribution

Format	Filename extension	Created by	Introduced in	Based on	Integrity check	Recovery record	Encryption supported	Unicode filenames	Modification date resolution
Debian package (deb)	.deb	Debian	1994 Debian 0.91	ar, tar, and gzip	Yes	No	No	Yes	1 s
Macintosh Installer	.pkg, .mpkg (metapackage)	NeXT	1989 NeXTSTEP 1.0	pax and gzip	Yes	?	?	Yes	?
RPM Package Manager (RPM)	.rpm	Red Hat	1995 Red Hat Linux 1.0	cpio and gzip	Yes	?	?	?	1 s
Slackware Package	.tgz	Patrick Volkerding	1993 Slackware 1.0	tar and gzip	Yes	No	No	?	?
Windows Installer (also MSI)	.msi	Microsoft	2000 Windows 2000	OLE Structured Storage, Cabinet and SQL	Optional PKCS7 Authenticode Signature	No	No	No	2 s
Java Archive (JAR)	.jar	Sun Microsystems	1997 JDK 1.1	PKZIP	Yes	?	?	Yes	?
Google Chrome extension package	.crx	Google	2009 (Chrome 4.0)	Zip	?	?	Yes	?	?
Pacman	.pkg.tar.xz	Judd Vinet	?	.tar.xz	?	?	?	?	?

Notes

^1 Not to be confused with the archiver JAR written by Robert K. Jung, which produces ".j" files.

Features

Archive format	Built-in compression	Self-extracting	Directory Structure	POSIX attributes	ACLs	Alternate data streams
cpio	No	No	Yes	Yes	?	?
tar	No	No	Yes	Yes	Yes	No
dar	Yes	No	Yes	Yes	Yes	Yes
ar	No	No	No	Yes	No	?
pax	No	No	Yes	Yes	Yes	?
dump	No	No	Yes	Yes	Yes	?
shar	No	Yes	Yes	Yes	?	?
makeself	Yes	Yes	Yes	Yes	Yes	?
zip	Yes	Yes	Yes	No	?	?
rar	Yes	Yes	Yes	No	?	?
ace	Yes	?	Yes	No	?	?
arj	Yes	Yes	Yes	No	No	?
zoo	Yes	?	Yes	No	?	?
ISO 9660 (CD-ROM)	No	No	Yes	(with Rock Ridge extension)	No	?
cab	Yes	Yes	?	No	?	?
rpm	Yes	No	Yes	Yes	?	?
deb	Yes	No	Yes	Yes	?	?
7z	Yes	No	Yes	Yes	?	?
Archive format	Built-in compression	Self-extracting	Directory Structure	POSIX attributes	ACLs	Alternate data streams

Notes

^1 Compression is not a built-in feature of the formats, however, the resulting archive can be compressed with any algorithm of choice. Several implementations include functionality to do this automatically
^2 That is, most implementations can optionally produce a self-extracting executable
^3 Per-file compression with gzip, bzip2, lzo, xz, lzma (as opposed to compressing the whole archive). An individual can choose not to compress already compressed filenames based on their suffix as well.

See also

• List of archive formats
• Comparison of file archivers
• Comparison of file systems
• List of file systems