Splunk Search

Histogram of transaction durations

dbcase
Motivator

Hi,

I have this query that finds the duration of the transaction times.

index=wholesale_app buildTarget=* product=* analyticType=sessionStart OR (analyticType=AppStateEvent AND Properties.index=3)|transaction clientSessionId startswith="sessionStart" endswith="AppStateEvent"|eval duration=if(duration<=50000,duration,0)|stats values(duration)

All well and good but what I'd like to do is have a count of how many transactions were in a given time period (range)

Something like this
0-10 seconds 4 transactions
11-30 seconds 2 transactions
31-60 seconds 1 transaction
1-3 minutes 8 transactions
3-10 minutes 21 transactions
etc etc

Something that looks like the below. Any thoughts?

alt text

0 Karma
1 Solution

cmerriman
Super Champion

you're going to have to do an eval. if the spans were the same, i'd say you could use bucket and do |bucket duration as bucket span=30s|stats count by bucket for instance, but that isn't the case.

something like this should work, just finish the case statement:
|eval bucket=case(duration<11,"0-10 seconds",duration<31,"11-30 seconds",duration<61,"31-60 seconds",duration<181,"1-3 minutes",duration<601,"3-10 minutes",.....,1=1,"Other")|stats count by bucket

you will also need to do an extra eval for each 0-10 seconds, 11-30 seconds.... to give them numerical values in order to sort them in the right order.
|stats count by bucket|eval sort=case(bucket="0-10 seconds",1,bucket="11-30 seconds",2,bucket="31-60 seconds",duration<181,"1-3 minutes",3,bucket="3-10 minutes",4,.....bucket="Other",x)|sort + sort|fields - sort

View solution in original post

cmerriman
Super Champion

you're going to have to do an eval. if the spans were the same, i'd say you could use bucket and do |bucket duration as bucket span=30s|stats count by bucket for instance, but that isn't the case.

something like this should work, just finish the case statement:
|eval bucket=case(duration<11,"0-10 seconds",duration<31,"11-30 seconds",duration<61,"31-60 seconds",duration<181,"1-3 minutes",duration<601,"3-10 minutes",.....,1=1,"Other")|stats count by bucket

you will also need to do an extra eval for each 0-10 seconds, 11-30 seconds.... to give them numerical values in order to sort them in the right order.
|stats count by bucket|eval sort=case(bucket="0-10 seconds",1,bucket="11-30 seconds",2,bucket="31-60 seconds",duration<181,"1-3 minutes",3,bucket="3-10 minutes",4,.....bucket="Other",x)|sort + sort|fields - sort

dragan_dragan
New Member

The sort can be simplified as

| eval bucket=case(
duration < 5,"0-4 seconds",
duration < 10,"5-9 seconds",
duration < 30,"11-30 seconds",
duration < 60,"30-59 seconds",
duration < 120,"1-2 minutes",
duration < 300,"2-5 minutes",
duration < 600,"5-10 minutes",
duration < 1200,"10-20 minutes",
1=1,"Other"
)
| stats max(duration) as maxDur, count by bucket
| sort +maxDur

0 Karma

dbcase
Motivator

Hi Cmerriman,

Many thanks! This is what I ended up with

index=wholesale_app buildTarget=* product=* analyticType=sessionStart OR (analyticType=AppStateEvent AND Properties.index=3)|transaction clientSessionId startswith="sessionStart" endswith="AppStateEvent"|eval duration=if(duration<=50000,duration,0)|eval bucket=case(duration<11,"0-10 seconds",duration<31,"11-30 seconds",duration<61,"31-60 seconds",duration<181,"1-3 minutes",duration<601,"4-10 minutes",duration<1801,"11-30 minutes",duration<3601,"31-60 minutes",duration>3600,"Greater than 1 hr",1=1,"Other")|stats count by bucket
0 Karma

dbcase
Motivator

Updated further

index=wholesale_app buildTarget=* product=* analyticType=sessionStart OR (analyticType=AppStateEvent AND Properties.index=3)|transaction clientSessionId startswith="sessionStart" endswith="AppStateEvent"|eval duration=if(duration<=50000,duration,0)|eval bucket=case(duration<11,"0-10 seconds",duration<31,"11-30 seconds",duration<61,"31-60 seconds",duration<181,"1-3 minutes",duration<601,"4-10 minutes",duration<1801,"11-30 minutes",duration<3601,"31-60 minutes",duration>3600,"Greater than 1 hr",1=1,"Other")|stats count by bucket|eval sorta=case(bucket=="0-10 seconds",1,bucket=="11-30 seconds",2,bucket=="31-60 seconds",3,bucket=="1-3 minutes",4,bucket=="4-10 minutes",5,bucket=="11-30 minutes",6,bucket=="31-60 minutes",7,bucket=="Greater than 1 hr",8,bucket=="Other",x)|sort + sorta|fields - sorta
0 Karma
Get Updates on the Splunk Community!

Easily Improve Agent Saturation with the Splunk Add-on for OpenTelemetry Collector

Agent Saturation What and Whys In application performance monitoring, saturation is defined as the total load ...

Explore the Latest Educational Offerings from Splunk [January 2025 Updates]

At Splunk Education, we are committed to providing a robust learning experience for all users, regardless of ...

Developer Spotlight with Paul Stout

Welcome to our very first developer spotlight release series where we'll feature some awesome Splunk ...