Open concaf opened 6 years ago
The most important part of the document should be what is the main problem that each of those projects is trying to solve.
Each project is looking at the application definition problem slightly differently. What I would like to see is what problem in that space is each project trying to solve with what approach.
My suspicion is that each project is trying to solve a different subset of the main problem.
One of the possible ways to showing differences might be to take some example application (ideally more than a single container, but not too complicated also :-) ) and then use each project, and document problems and benefits. (It should show initial deployment and also updating existing deployment)
Should we also consider ksonnet for comparsion?
I think we should. @kadel
I'd like to assign this to myself.
I'm trying to make this into a table. Anyone have any suggestions?
Kompose | Ksonnet | OpenShift templates | Kedge | |
---|---|---|---|---|
Scaleable | - | ✓ | ✓ | ✓ |
Native to Kubernetes | - | ✓ | - | ✓ |
Integration to tools (IDEs) | ✓ | ✓ | ✓ | ✓ |
YAML Formatted | ✓ | - | ✓ | ✓ |
Parameterized | ✓ | ✓ | ✓ | ✓ |
@containscafeine @surajnarwade @kadel @pradeepto
It looks good @cdrage. Helm should be also added to the comparison. Comparison Helm vs Kedge was brought in various discussions.
@kadel Woops. I completely forgot about Helm. I'll add it and push this to a PR.
PR here: #571
Umm, this shows Kedge as a competitor to the other tools, while it does not necessarily overlap with Helm, Templates, etc.
@containscafeine Do you have any suggestions regarding comparing them? Yeah.. it was a bit difficult compare them all in one simple diagram / image. But I came up with the table :man_shrugging: Should we remove some of the other references since they're not necessarily competitors (I think Kedge is the only one in it's niche)
@cdrage sorry for the vague reply. So, I agree with the table mostly, it's not factually incorrect, it's just that folks keep on coming up every now and then comparing kedge to other tools like Helm, etc, which is unfair because the tools are not even in the same domain. If we put this comparison table, then we send a message that yes, kedge is actually comaprable to those tools, which is wrong. We should probably have something like https://github.com/kubernetes-incubator/kubespray/blob/master/docs/comparisons.md, which explains in detail what's wrong with the comparisons and under which use case, does the said tool fall.
@kadel suggested and I agree that it'd really help to have a comparison doc with other projects in the same space and point out the difference in how is Kedge different in approach towards solving the application definition problem.
The tone of the doc should be only to point out the differences and comparing technical approaches.
Some projects I can think of that are generally thought of as overlapping with Kedge are - Helm, Kompose, APB, Kompose, OpenShift templates, etc