Database Reference
In-Depth Information
Monitoring
Systems monitoring is the classic application for real-time data collection
and analysis. This usually involves collecting the number of events over a
period of time quantized to a particular frequency.
For example, a website might collect the number of each type of HTTP
responses per minute. Changes in the frequency of error (5xx) responses
usually warrant some type of response, although they also occur due to
transitory events such as extreme load spikes. A change in the frequency
of correct (2xx) responses might also warrant a response. For a news site,
a breaking story that “goes viral” could have any number of possible
responses.
There are also the classic system-monitoring applications that track aspects
of the machines running the website. In this case, different gauges are
tracked:CPUload,fanspeed,diskandnetworkI/O.Changesinthesevalues
are often indicative of external changes or failing hardware. Other more
fanciful applications might include large-scale medical monitoring for
hospitals or traffic monitoring for freeways.
In any case, irrespective of the application area, the goal of monitoring
systems is simple: Identify changes to the system, notify operators of the
