Syslog, the default logging mechanism on Unix and Linux, dates from the early 1980s and was originally developed by Eric Allman – initially for Sendmail. Later, the tool established itself as a universal solution for logging system and error messages of all kinds. Among Linux distributions, Syslog was the generally accepted standard for many years. But while Syslog was marching to triumph, it revealed a number of weaknesses:

• The protocol does not provide authentication; anyone can generate spoofed log entries for any application.
• Syslogd transmits all messages in plain text, and anyone can read them.
• The timestamp does not contain any information about the time zone.
• As its transport protocol, Syslog uses the connectionless UDP, which does not guarantee that all messages arrive.
• Browsing the log files is a relatively complicated process, requiring tools that search for text patterns.
• The metadata that the Syslog protocol stores is incomplete.
• You can't log binary data.
• Syslog is only one of several logs on Linux, and the user must separately access UTMP/WTMP, Lastlog, audit, kernel logs, and firmware logs, as well as a variety of application-specific logs.
• Log rotation and compression are available but not flexible. Rotation only applies to fixed intervals but does not include disk utilization, and compression usually only works during rotation.

The Syslog protocol was first standardized in 2001 in RFC 3164. Developers soon created alternatives that corrected some of the weaknesses, including Syslog-ng and Rsyslog. Syslog-ng supports dozens of features that go beyond the old Syslog daemon's feature set; however, some of these features – including encryption, multi-line messages, and failover on the client side – only exist in a commercial premium edition.

[...]