I did work through the snippet@Jay's repo on my end in using Kubespray[1]
and vagrant VM. There are minor nits but overall it works well as
expected...
In my opinion, it's breaking changes for us because the tooling that we've
used is not integrated(or replaced) yet. so I think it would be good to
start with a development branch based on Jay's work... and also new branch
is harmless on current development.

I'm proposing a new branch for **incubating** cloud-native distribution
which is based on Jay's ground work!

What do you think?

Thanks,
Youngwoo

1. https://github.com/kubernetes-sigs/kubespray

On Mon, Sep 16, 2019 at 8:02 PM Jay Vyas <[email protected]>
wrote:

> - No need for our own k8s.  Let’s just test with nfs volume provisioned
> and note what version of nfs, networking, helm,  and cluster were using to
> test releases .
>
> - I think since we are using helm charts , we can add anything and curate
> and test.   Cost of packaging with helm and docker  is very low.
>
> - If you go to the repo there is a 20 line snippet we can start with on
> the readme that sets up Kafka spark NiFi zk minio and presto.
>
> > On Sep 15, 2019, at 10:10 PM, Youngwoo Kim (김영우) <[email protected]>
> wrote:
> >
> > Good job Jay & Sid. Nice work! very informative and also thanks Jay
> > for sharing the stuffs.
> >
> > I have a couple of questions and suggestions:
> > - Do we need our own k8s distribution? (pros vs. cons)
> > - k8s is complex and also people love (easy)UI :-) How can we reduce the
> > gap?
> > - How can we test the distribution? (integrity and performance?)
> > - Rest of the elements?, Such as logging, metrics & security.
> > - How do we manage base container images? official images or github?
> >
> > As a distribution for big data and analytics, it would be nice to have
> such
> > components(already... there are nice projects in ASF :-)):
> > - Jupyter Notebook and jupyterhub
> > - More choices on stream data processing: Flink, Pulsar
> > - Task scheduling: Airflow
> > - Interactive analytics and dashboard: Superset
> > - Functions(serverless): OpenWhisk or KNative
> > - Purpose-built data store for big data: Redis, Cassandra and Druid.
> > - and more...
> >
> > WDYT?
> >
> > Thanks,
> > Youngwoo
> >
> > On Sun, Sep 15, 2019 at 2:07 AM jay vyas <[email protected]>
> > wrote:
> >
> >> Hi everyone !
> >>
> >> The contents for this that were finally presented are here...
> >> https://github.com/jayunit100/apachecon-2019-bigtop3x
> >>
> >> and the video is here !  https://youtu.be/LUCE63qPMyM
> >>
> >> Let us know what you think
> >>
> >>
> >> On Thu, Sep 12, 2019 at 4:35 PM jay vyas <[email protected]>
> >> wrote:
> >>
> >>> hi folks, am in the conference lobby area now.  have 2 laptops out so
> >> easy
> >>> to find :) .  .
> >>>
> >>>
> >>> On Thu, Sep 12, 2019 at 2:41 PM Konstantin Boudnik <[email protected]>
> >> wrote:
> >>>
> >>>> Well, here's an interesting piece of news:
> >>>>
> >>>>
> >>>>
> >>
> https://www.theregister.co.uk/2019/09/10/google_cloud_lights_spark_for_kubernetes/
> >>>>
> >>>> Mind you, these nice fellas [/sarc] promised on two different
> occasions
> >>>> that
> >>>> they will be contributing back to Apache Bigtop - the foundation of
> the
> >>>> Dataproc - and yet they never did.  But I guess this is fine ... not
> the
> >>>> first
> >>>> company taking this kind of advantage :)
> >>>>
> >>>> --
> >>>>  Cos
> >>>>
> >>>>
> >>>>> On Wed, Sep 04, 2019 at 05:54AM, Jay Vyas wrote:
> >>>>> Hi folks.  Couple of updateys  in the next gen brainstorming we’ve
> >> been
> >>>> doing...
> >>>>>
> >>>>> Sid Mani from minio is going to copresent with me at our bigtop talk
> >> at
> >>>>> apachecon next week, around the internals of minio integrations etc.
> >>>>>
> >>>>> Spoke with cos and sounds like we definetly have a quorum for hacking
> >>>> on this next gen implementation.
> >>>>>
> >>>>> Whose in on hacking on it,  and who will be in Vegas on the 12th?
> >>>>>
> >>>>> I’ll help w the k8s stuff so don’t be intimidated if that’s your weak
> >>>> spot.
> >>>>
> >>>
> >>>
> >>> --
> >>> jay vyas
> >>>
> >>
> >>
> >> --
> >> jay vyas
> >>
>

Reply via email to