Build a chart of multiple data series
Splunk transforming commands do not support a direct way to define multiple data series in your charts (or timecharts). However, you CAN achieve this using a combination of the
timechart commands both return tabulated data for graphing, where the x-axis is either some arbitrary field or
_time, respectively. When these commands are used with a split-by field, the output is a table where each column represents a distinct value of the split-by field.
In contrast, the
stats command produces a table where each row represents a single unique combination of the values of the group-by fields. You can then use the
xyseries command to redefine your data series for graphing.
For most cases, you can simulate the results of "... | chart n by x,y" with "... | stats n by x,y | xyseries x y n". (For the
timechart equivalent of results, x =
Let's say you want to report on data from a cluster of application servers. The events gathered from each server contain information such as counts of active sessions, requests handled since last update, etc. and are placed in the
applications_servers index. You want to display each server instance and the number of sessions per instance on the same timechart so that you can compare the distributions of sessions and load.
Ideally, you want to be able to run a timechart report, such as:
| timechart sum(handledRequests) avg(sessions) by source
However, timechart does not support multiple data series; so instead, you need run a search similar to the following:
| bin _time
| stats sum(handledRequests) as hRs, avg(sessions) as ssns by _time, source
| eval s1="handledReqs sessions"
| makemv s1
| mvexpand s1
| eval yval=case(s1=="handledReqs",hRs,s1=="sessions",ssns)
| eval series=source+":"+s1
| xyseries _time,series,yval
Let's break this search down so it's easier to understand what each part of the search is doing:
The first thing that you need to do, before the
stats command, is to separate the events by time.
... | bin _time
stats command is used to calculate statistics for each source value: The sum of
handledRequests values are renamed as
hRs, and the average number of
sessions are renamed as
... | stats sum(handledRequests) as hRs, avg(sessions) as ssns by _time, source
The following portion of the search uses the
eval command to add a single-valued field called "s1" to each result from the
stats command. Then, the
makemv command converts the values in the
s1 field into a multivalued field, where the first value is "handledRequests" and the second value is "sessions". The
mvexpand command then creates separate series for each value of s1.
... | eval s1="handledRequests sessions" | makemv s1 | mvexpand s1
Then the search uses the
eval command to define a new field called "yval", and assign values to the field based on the case that it matches. So, if the value of s1 is "handledRequests", the
yval field is assigned the "hRs" value. And, if the value of the
s1 field is "sessions", the
yval field is assigned the "ssns" value.
... | eval yval=case(s1=="handledRequests",hRs,s1=="sessions",ssns)
Then the search uses the
eval command to define a new field called "series", which concatenates the value of the
... | eval series=source+":"+s1
xyseries command is used to define a chart with
_time on the x-axis,
yval on the y-axis, and data defined by the
... | xyseries _time, yval, series
- bin command in the Search Reference
- eval command in the Search Reference
- makemv command in the Search Reference
- mvexpand command in the Search Reference
- stats command in the Search Reference
- xyseries command in the Search Reference
Look for associations, statistical correlations, and differences in search results
Compare hourly sums across multiple days
This documentation applies to the following versions of Splunk Cloud Platform™: 8.2.2112, 8.2.2201, 8.2.2202, 8.2.2203, 9.0.2205, 9.0.2208, 9.0.2209 (latest FedRAMP release), 9.0.2303
Feedback submitted, thanks!