| HAProxy | |
|---|---|
 | |
| Original author | Willy Tarreau |
| Initial release | December 16, 2001; 24 years ago (2001-12-16) |
| Stable release | 3.2.0[1]  / 28 May 2025 |
| Written in | C |
| Operating system | Linux,FreeBSD,OpenBSD,Solaris (8/9/10),AIX (5.1–5.3),macOS,Cygwin |
| License | GNU General Public License Version 2 |
| Website | www |
| Repository | |
HAProxy is afree and open source software that provides ahigh availabilityload balancer and proxy (forward proxy,[2]reverse proxy) forTCP andHTTP-based applications that spreads requests across multiple servers.[3] It is written inC[4] and has a reputation for being fast and efficient (in terms of processor and memory usage).[5]
HAProxy is used by a number of high-profile websites includingGoDaddy,GitHub,Bitbucket,[6]Stack Overflow,[7]Reddit,Slack,[8]Speedtest.net,Tumblr,Twitter[9][10] andTuenti[11] and is used in theOpsWorks product fromAmazon Web Services.[12]
HAProxy was written in 2000[13] by Willy Tarreau,[14] a core contributor to theLinux kernel,[15] who still maintains the project.
In 2013, the company HAProxy Technologies, LLC was created.[citation needed] The company provides a commercial offering, HAProxy Enterprise and appliance-basedapplication-delivery controllers named ALOHA.
HAProxy has the following features:
HAProxy Enterprise Edition is an enterprise-class version of HAProxy that includes enterprise suite of add-ons, expert support, and professional services. It has some features backported from the HAProxy development branch.[20]
HAProxy Technologies’ ALOHA is aplug-and-play load-balancing appliance that can be deployed in any environment. ALOHA provides agraphical interface and a templating system that can be used to deploy and configure the appliance.[21]
HAProxy has had the following version releases:[22]
| Version | Release date | End of life |
|---|---|---|
| Unsupported: 1.0 | 2001-12-16 | 2001-12-30 |
| Unsupported: 1.1 | 2002-03-10 | 2006-01-29 |
| Unsupported: 1.2 | 2003-11-09 | 2011-08-06 |
| Unsupported: 1.3 | 2006-06-29 | 2016-03-14 |
| Unsupported: 1.4 | 2010-02-26 | 2018-02-08 |
| Unsupported: 1.5 | 2014-06-19 | 2020-01-10 |
| Unsupported: 1.6 | 2015-10-13 | 2020-Q4 |
| Unsupported: 1.7 | 2016-11-25 | 2021-Q4 |
| Unsupported: 1.8 | 2017-11-26 | 2022-Q4 |
| Unsupported: 1.9 | 2018-12-19 | 2020-Q2 |
| Unsupported: 2.0 | 2019-06-16 | 2024-Q2 |
| Unsupported: 2.1 | 2019-11-25 | 2021-Q1 |
| Supported: 2.2 LTS | 2020-07-07 | 2025-Q2 (critical fixes only) |
| Unsupported: 2.3 | 2020-11-05 | 2022-Q1 |
| Supported: 2.4 LTS | 2021-05-14 | 2026-Q2 (critical fixes only) |
| Unsupported: 2.5 | 2021-11-23 | 2023-Q1 |
| Supported: 2.6 LTS | 2022-05-31 | 2027-Q2 |
| Unsupported: 2.7 | 2022-12-01 | 2024-Q1 |
| Supported: 2.8 LTS | 2023-05-31 | 2028-Q2 |
| Unsupported: 2.9 | 2023-12-05 | 2025-Q1 |
| Supported: 3.0 LTS | 2024-05-29 | 2029-Q2 |
| Supported: 3.1 | 2024-11-26 | 2026-Q1 |
| Latest version:3.2 LTS | 2025-05-28 | 2030-Q2 |
Legend: Unsupported Supported Latest version Preview version Future version | ||
Servers equipped with 6 to 8 cores generally achieve between 200,000 and 500,000 requests per second, and have no trouble saturating a 25 Gbit/s connection under Linux.[23] 64-core ARM servers were shown to reach 2 million requests per second and 100 Gbit/s.[24]
