cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Timeseries API Aggregation

bill_scheuernst
DynaMight
DynaMight

Still working on the migration from SolarWinds (SW) to Dynatrace (DT) for Cisco network equipment.  Currently we ingest data for vpn sessions from SW into Splunk and calculate number of sessions for various timeframes and for each data center.  In DT, and using a custom version of the Cisco SNMP extension, we are injesting data into DT for the same device type, and would like to have a similar visualization and calculation for vpn sessions.  This is regarding a metric in the SNMP extension for ASA crasNumSessions metric.  The extension we are using polls each device every 3 minutes, as opposed to the standard DT poll of 1 minute.  When performing a sum calculation on the results, we are seeing results different from what is showing in Splunk and we believe the problem lies with different datasets due to the polling frequency differences.

Support Ticket - 318810

Has anyone encountered this situation before for session counts?  How do you adjust for the difference in polling frequency?  The Grail output shows null values where normal polling data of 1 minute would be available.  We would like to get a total sum of these sessions for the VPN devices, but not split by device.  When splitting by device and select "Last Value" as the fold transformation, the values are correct, however when removing the split, the total session count is incorrect.  We believe the polling frequency is causing an issue when changing the default DT time resolution of Auto, 1min, or 5min.  The time resolution change displays the wrong values.  The devices do not poll all at the same 3 minute interval.

The attached documents better explains the situation.

Dynatrace Certified Professional
1 REPLY 1

ChadTurner
DynaMight Legend
DynaMight Legend

in terms of alerting, we put in a latency value. So if it takes 2 mins for the data to show, we say -2m, which then prevents the system from seeing and alerting on null data. Id be curious to see the data its collecting. 

If you are polling every 3mins, I would set the latency as such. And when it polls, lets say 8:00 you have data, then 8:03, you get the next set of data. Is that data point back filling 8:01, 8:02, and 8:03? or is it just 8:00, 8:03, 8:06 data points that are coming up? 

-Chad

Featured Posts