Getting Data In

How can we speed up JSON array lookup?

vinodkd
New Member

Hi,

My JSON event is in this form:

{
TotalMemUsage : 887992,
ProcMemUsage : [
{
Name : "firefox",
PID : 758,
Usg : 228972,
},
{
Name : "eclipse",
PID : 569,
Usg : 19756,
}
]
}

I've to take average of each process's memory usage and draw a chart.
For the time being, I use following query.

my_search 
| rename ProcMemUsage{}.Name as PNAME | rename ProcMemUsage{}.Usg as PUSAGE
| eval x=mvzip(PNAME,PUSAGE) 
| mvexpand x|eval x=split(x,",")
| eval P_NAME=mvindex(x,0)
| eval P_USAGE = mvindex(x,1)
| stats avg(P_USAGE) as MU by P_NAME 
| sort -MU | fields MU,P_NAME
|chart avg(MU) as AvgMU by P_NAME 
| sort -AvgMU

But it takes lot of time to complete the operation. (Approx 5 minutes with only 30K records.).
Is there any way to optimize it? Can we use jsonutils somehow?

Tags (3)
0 Karma

vinodkd
New Member

@Iguinn: It doesn't give a noticeable speedup :'(

0 Karma

lguinn2
Legend

I don't really think this will make things faster, but it might help a little:

my_search 
| rename ProcMemUsage{}.Name as PNAME | rename ProcMemUsage{}.Usg as PUSAGE
| fields PNAME PUSAGE
| eval x=mvzip(PNAME,PUSAGE) 
| mvexpand x
| eval x=split(x,",")
| eval P_NAME=mvindex(x,0)
| eval P_USAGE = mvindex(x,1)
| chart avg(P_USAGE) as AvgMU by P_NAME 
| sort -AvgMU

It's also a bit cleaner. Post back with your results.

0 Karma
Get Updates on the Splunk Community!

Community Content Calendar, August edition

In the dynamic world of cybersecurity, staying ahead means constantly solving new puzzles and optimizing your ...

Pro Tips for First-Time .conf Attendees: Advice from SplunkTrust

Heading to your first .Conf? You’re in for an unforgettable ride — learning, networking, swag collecting, ...

Introducing Splunk 10.0: Smarter, Faster, and More Powerful Than Ever

Whether you're managing complex deployments or looking to future-proof your data infrastructure, this session ...