The term “monitoring” needs clarification.

WebOps-related mailing lists have always had a problem with this vague term, and I suspect that commercial vendors exploit this confusion. 🙂

Wikipedia gives a pretty vague definition:

“…is the process of testing or tracking (monitoring) how end-users interact with a website or web application.”

People use the term to describe lots of things that pertain to the web:

  1. Global availability and simulated end-user performance (like Gomez or Keynote)
  2. Server and network health checking (like nagios, mon, or Big Brother, etc.)
  3. Network and server statistic graphing/trend tools (like munin, ganglia, cacti, etc.)

I wish I knew of a better way to disambiguate the term in an easy way. I suspect that the tools listed above will always use the word ‘monitoring’ in their descriptions. Sigh.

Relatedly, I have the opinion that the three different functions listed above should be performed by different pieces. I’ve yet to see any tool that sucessfully combines health-check/alerting with resource trend/graphing, for example. I don’t have much experience with GroundWork or Zabbix, maybe folks who do can argue with me on that point. 🙂


  1. sysadmin@pt   •  

    Indeed. I usually use the term monitoring when referring to server and netwrok health checking but I also see the term applied to the other options. Most people I know use it do define graphing tools.

  2. Peter Mui   •  

    The BayLISA Monitoring SIG, which has been meeting for over a year now, classified Monitoring functionality into these four areas:


    It isn’t perfect, but it was the best we could do after much discussion and deliberation and hemming and hawing. (For example, do “Aggregation” and “Notification” deserve separate mention, or are they functionally a subset of one of the above?)

    So one of the upcoming goals of the SIG is to map popular monitoring tools against these classifications. Obviously, some straddle more than one.

Comments are closed.