"CORE" namespace in XDM identityMap

[jgrubbs]

Hi gang,

I have what I hope is a simple question for someone who has Edge and Konductor inside secrets.

We are currently in a RTCDP/AEM implementation project with a team of focused Adobe consultants. As they validate our data sources and our stream of events via the WebSDK, they are seeing hits where the xdm.identityMap is missing the ECID namespace, but instead has a namespace called "CORE" which does have what appears to be a legit ECID set in the id field. It is a 38 character numeric expression.

Obviously our client side code isn't setting that, so my hunch is that it's something that Konductor is responsible for. Other clues are that these hits are likely coming from clients with no pre-existing cookie or ECID, so my theory is that when Konductor has to reach out to the ID service to get an ECID, it is placing it there in "CORE" and not "ECID"?

Can someone shed some light on this little mystery? You'd make my Friday!

[pradeep_nextrow · Accepted solution]

@jgrubbs  "CORE" is the namespace for AAM identifier (I think this is the UUID and that is the reason for the mix up). The konductor (Edge) sets the ECID server side and also ECID lives (I think in an encrypted format) in the konductor core identity cookie. In every request to Edge, you will see a "query" under XDM, where alloy goes and asks for the ECID. You can actually see if this would be resulting in a successful ECID in the response ("Preview" tab of the Chrome debugger). Here you should get the ECID along with the namespace under "identity:result" like in the screenshot. (Please do this as a new visitor from clean browser window).

You can also see this with the help of Adobe Experience Platform Debugger:

Bottom line, you should see the the ECID under the right namespace and not under "CORE". Check your Data Stream settings for Audience Manager, if any, disable and try once. Also check your Web SDK (alloy.js) configuration in Adobe Launch and make sure you have identity migration enabled.

In your debugger, you can also see a validation of the call. If that goes through, then you should have no problems in ingesting the data/batches. My sandbox is down, so I am not able to send you a screenshot now, will send once it is back.

[jgrubbs]

@pradeep_nextrowAh cool that answers my question about the elusive CORE namespace   The format does match the legacy aam_uuid as well, which is also a 38 character numeric string

I wrote some tools to retrieve the ingested batch files from AEM via the API in parquet format and wrote a parser to check all the records. Konductor is indeed setting the ECID namespace in the records as I expected it would. However, it was those handful of records with "CORE" that were a mystery... until now    I wish details like this were documented (maybe it is, but I was never able to find any), but I guess that's why this community is so awesome!