4 ms·
I'd be interested to know what important features added to syslog-ng that the author is talking about. I'd consider myself to be in the "syslog-ng is just a dr
by 22c 5y ago
I'd be interested to know what important features added to syslog-ng that the author is talking about.
I'd consider myself to be in the "syslog-ng is just a drop-in replacement for syslogd, right?" camp, so I'm certainly interested in hearing about what neat things have been added already. Perhaps it would help some people to learn about some of those lesser known features before they feel comfortable making suggestions for even more features.
- bazsi77 5y agoI was told the other day the biggest limitation of syslog-ng is its name. Everyone assumes it is about "syslog", while it has been able to do much more than syslog for a long time. It is a drop-in replacement of syslogd, and it is true that syslog is a pretty important use-case, but here's a list of features I consider "extra". I plan to write a more detailed blog post later. High level capabilities: - curating and fixing up the incoming message stream, so you can use it easier in your SIEM or other downstream log consumer - application classification, e.g. detect what app is sending a specific message, use this logic to determine sourcetype in Splunk for instance - native code, small footprint (CPU, memory), scale over 100k/sec easily, some use-cases even 600-800k/sec - powerful config language, allowing all of the above. Not coding, but configuration (e.g not javascript, typescript, ruby or Java snippets) - but if the config language is not enough, easily extendable using Python (or Java) Features: - generic key-value pairs, actually each message is a set of key-value pairs - JSON support, we parse & produce and translate JSON documents - HTTP destination (integration with Splunk for instance, but any HTTP API is doable, see our Twitter/Discord integrations) - stream based correlation (e.g transform an event spanning multiple messages into 1 aggregated event) - Kafka for enterprise queueing - MQTT for IoT and sensor data - Elastic/MongoDB/SQL integration for databases - Riemann integration for monitoring I am afraid this is becoming long and some details could be added for each of the above. I make an attempt to cover these areas in my subsequent blog posts. So consider subscribing if any of these is interesting.