topic Re: Comparing information based on _time and calculated value in Splunk Search

Comparing information based on _time and calculated value

Brainstorms — Thu, 30 Dec 2021 17:52:46 GMT

I have looked for solutions but I have mostly found results regarding only current and past time comparison which is not what I need.

I have a query that bins _time by 24h spans over the previous 7 days. and calculates a numeric value associated with those time spans. What I need is to compare each day's values to the entire week's and find any time period (so 48h) where the number jumped significantly.

An example of something similar to my my code:

index=sandwiches saved_search_name="yum" earliest=-7d
| bin span=24h _time
| search sandwich_type="PB&J"
| stats count by total_bread_type _time
| stats sum(total_bread_type) as bread by _time
| eval bread = round(bread / 10000, 2)

currently the results are like this:

_time	bread
2021-12-22 18:00	22
2021-12-23 18:00	23
2021-12-24 18:00	21
2021-12-25 18:00	47
2021-12-26 18:00	48
2021-12-27 18:00	46
2021-12-28 18:00	47

Basically I am looking to compare the 'bread' values by _time and figure out if/where there is a jump of 10 or more and return that data.

Any insight would be appreciated. Thanks!

Re: Comparing information based on _time and calculated value

richgalloway — Thu, 30 Dec 2021 20:22:58 GMT

Try this. The streamstats command takes the difference between adjacent results then we filter for a jump of more than 10.

Re: Comparing information based on _time and calculated value

Brainstorms — Thu, 30 Dec 2021 20:26:12 GMT

That did the trick! Thank you so much Rich.