Information Technology Reference
In-Depth Information
Terms to Know
Server:
Software running to provide a function or API. (Not a piece of hard-
ware.)
Service:
A user-visible system or product composed of many servers.
Machine:
A virtual or physical machine.
QPS:
Queries per second. Usually how many web hits or API calls received per
second.
Diurnal Cycles:
Metrics that are high during the day and low at night.
Instead, monitoring should be designed to detect the precursors of an outage in time for
theproblemtobeprevented.Asystemmustbeinstrumentedandmonitoredsoastoenable
this strategy. This is more difficult than detecting an outage, but much better.
16.1 Overview
To understanding monitoring you must first understand its particular terminology.
A
measurement
referstoasinglepointofdatadescribinganaspectofasystem,usually
a value on which numerical operations make sense—for example, 5, -25, 0, or Null. It can
also be a string—for example, a version number or a comma-separated list of currently
mounted file systems.
A
metric
is a measurement with a name and timestamp. For example:
spellcheck:server5.demo.com:request-
count@20140214T100000Z = 9566
Different monitoring systems use different formats to label the data. In the preceding
example, we have a metric from the spellcheck server that was running on server5. The
measurement is the request-count, presumably how many requests the server has received
since starting. After the @ symbol is the date stamp. The measured value is 9566.
Examples of other metrics include the total RAM in use, the number of outstanding re-
quests in a queue, the build version of the code being run, the number of open bugs, the
number of bugs, and the total cost on Amazon AWS.
There is also metadata associated with a metric. In particular, a numeric value usually
has a unit associated with it. Knowing the unit permits automatic conversions, chart label-
ing,andsoon.Manymonitoringsystemsdonottracktheunitsofametric,requringpeople
Search WWH ::
Custom Search