Replies: 2 comments
-
There is on-going work in the ETSI-NFV SOL004 workgroup to address CNF packaging. There is also work in the ONAP community to develop the implementation of an orchestrator that deals with those packages (https://wiki.lfnetworking.org/pages/viewpage.action?pageId=50528918&preview=/50528918/50528919/LFN-DDTF-ONAP-ETSI-Alignment%20Architecture%20for%20Honolulu-Plus-2021-2-2.pptx). Even for those who are not big fans of ETSI-NFV, it is at least worth a look, just so the wheel is not re-invented here. |
Beta Was this translation helpful? Give feedback.
-
This has also to include the deployment or usage of private registries, specially for those cases where networks don't have internet access. |
Beta Was this translation helpful? Give feedback.
-
We need to ensure that a CNF, with all its bits, are deliverable. We assume that many CNFs will be provided by third parties. We need a consistent install and use experience across CNFs, to avoid exploding the ops workload because we install and start and use every CNF in a unique way. This begins by ensuring they're all wrapped up in the same way.
Transfer into its running environment may need to be sneakernet, since there are some networks that are explicitly and deliberately not internet connected.
Sub-discussion: I am specifically thinking that this should include some metadata as well as the software. I need something in that package:
Which Helm chart?)
Note in the above: I think we should be preserving the ability to change our minds about this stuff in the future. What I'm thinking here is that if the CNF documents what it does, even if there's only one choice today, it means that if we find a better way of doing this in the future we can write in the metadata that this uses the old method or the new one.
Beta Was this translation helpful? Give feedback.
All reactions