I think we should strive for a single set of Helm Charts. Configuration overrides for the target deployment environment would then be applied to enable/disable sub components and provide central or edge configuration values.

The target environments (central vs edge) may influence how the DMaaP microservices are created and assembled. Assuming that some services might not be necessary in one or more environments.

Mike.

From: "LUNANUOVA, DOMINIC (DOMINIC)" <dgl@research.att.com>
Date: Thursday, December 20, 2018 at 9:00 AM
To: Mike Elliott <Mike.Elliott@amdocs.com>, "GUAN, HONG" <hg4105@att.com>, "UNNAVA, SUNIL" <su622b@att.com>, Fiachra Corcoran <fiachra.corcoran@ericsson.com>, Conor Ward <conor.ward@ericsson.com>, Mark Scott <mark.scott@ericsson.com>
Cc: Tom O'Neill <tom.o.neill@ericsson.com>, "Choquette, Marc-Alexandre" <m.a.choquette@bell.ca>, "Landry, Marc" <Marc.Landry@bellaliant.ca>
Subject: RE: DMaaP Dublin Deployment Requirements

Mike,

During our meeting, the observation was made: “an ONAP deployment in an Edge K8S would likely be a subset of ONAP”.

I think our DMaaP examples bear this out, and there are likely other examples.

I anticipate that there may be components which only deploy at Edge too.

So, from a Helm Chart perspective, do you see a project like DMaaP maintaining a single Chart that has conditional logic  for whether it is an Edge deployment,  or do you think we should have separate Charts for Central and Edge?

-Dom