There appear to be two sets of images we're concerned with: Reference images 
(RI), and those images that will be built and deployed by operators in the real 
world.

I believe our intent is to define a build chain and rules or guidelines that 
all projects use to build their reference images. As a general model (hopefully 
not too simplistic), an RI is composed of the following 'logical' layers (not 
Docker layers):
1. architecture
2. OS
3. libraries (e.g Java, Python, etc)
4. application

Layers 1-3 can be 'standardized' for all ONAP projects. 

A few observations/questions:
* we want to support multi-arch images so all RI images are in practice 
Manifest lists (or equivalent in OCI) - right?
* do we standardize on one OS for the purposes of testing within ONAP? (I 
believe this is what Tal has suggested)
* The RI image repo should have images that contain static content for many 
application environments (layer 3 above), which ameliorates the concerns 
Micheal identified just above, shortens start-up times, and provides a more 
uniform baseline for applications - correct?
* Once we agree on concrete items on this thread, we should memorialize them on 
a wiki page.

Now for 'real world' images - presumably built from scratch by each operator. 
These images might not share layers 1-3, but will share layer 4, the ONAP 
component applications. What can we do to facilitate this build/test/deploy 
process to make life easier, safer, and less error prone for operators?

- Frank

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#11931): https://lists.onap.org/g/onap-discuss/message/11931
Mute This Topic: https://lists.onap.org/mt/24242182/21656
Group Owner: [email protected]
Unsubscribe: https://lists.onap.org/g/onap-discuss/unsub  
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to