edgent-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gayashan Amarasinghe <gayashan.amarasin...@gmail.com>
Subject Re: [DRAFT] November Podling Report Edgent
Date Wed, 07 Nov 2018 00:05:10 GMT
Hi Julian,

Thanks for pointing to NiFi, I'll have a look at it. I used edgent at that
time since it was simple enough to get a custom topology up and running
without any added bloat.

Regarding the dataset, it is a series of unrelated events and definitely
partitioning is possible. But I was not worried about that aspect, since I
was focusing on placement of the topology and not on improving the
performance through parallelism of some sort. Is it why you asked that?

I'll start a new thread on the fault tolerance aspects. It should be
interesting to understand others' view on it.

/Gayashan


On Tue, Nov 6, 2018 at 8:58 PM Julian Feinauer <j.feinauer@pragmaticminds.de>
wrote:

> Hey Gayashan,
>
> thanks for the insights.
> Sounds very interesting, another interesting tool for that could be Apache
> NiFi / MiNiFy which has this "partitioning already in its core (but better
> keep here ; )
> Are these datasets timeseries of related or rather unrelated events, so
> does order matter, is partitioning possible?
>
> And regarding Fault Tolerance, that is a wide and very interesting field,
> but I agree, better start a new thread for that and then discuss it there
> in dept.
>
> Julian
>
> Am 06.11.18, 10:51 schrieb "Gayashan Amarasinghe" <
> gayashan.amarasinghe@gmail.com>:
>
>     HI Julian,
>
>     Thank you for your kind reply and understanding.
>
>     I've been working on sharing the work load of distributed stream
> processing
>     between the edge and the cloud. So what I have been doing now is
> running
>     edgent on both the edge and the cloud. To be specific I was running a
> part
>     of the edgent streaming topology on the edge where my event sources are
>     located, do some processing there, send the processed stream to the
> cloud
>     and run the final part of the streaming topology on the cloud to
> complete
>     the processing there. So basically a big edgent topology partitioned
>     between the edge and the cloud. Does that make sense?
>
>     I'm currently looking at my next idea and thinking of fault tolerance
>     aspects that could be required. What are your thoughts on fault
> tolerance
>     requirements of edgent from an industry/real user perspective? (May be
> I
>     should bring this discussion to a new thread on the list.)
>
>     Thanks!
>
>     Best,
>     Gayashan
>
>
>     On Tue, Nov 6, 2018 at 6:47 PM Julian Feinauer <
> j.feinauer@pragmaticminds.de>
>     wrote:
>
>     > Hi Nuwan,
>     >
>     > it's a pleasure to hear from you!
>     > I didn’t know that there is so much usage of edge processing in
> academia,
>     > that’s pretty cool to hear.
>     > As I tried to dig into the Eclipse IoT Eco System the last time,
> it's good
>     > to hear that we have people on the list that are familiar with
> Mozillas
>     > Gateway!
>     >
>     > And, to be honest, it would be great if you could find some time to
>     > contribute documentation or media as this is something programmers
> tend to
>     > skip __
>     >
>     > Best
>     > Julian
>     >
>     > PS.:
>     > @Nuwan and @Gayashan:
>     > Would it be possible for you to share some details about your use
> cases
>     > (where you used edgent, or at least considered to use edgent) with
> us?
>     > I would be really interested in it.
>     >
>     > Am 06.11.18, 04:12 schrieb "Nuwan Jayawardene" <nsgaj12@gmail.com>:
>     >
>     >     Hi All,
>     >
>     >     I'm an undergrad following a CS degree in Sri Lanka.
>     >     I came across Edgent when doing research for my final year
> thesis on
>     > Fog
>     >     Networks and IoT.
>     >     Even though Edgent isn't in my in project scope (it was
> overkill, I
>     > went
>     >     with Mozilla's Things Gateway) for the time being, I had definite
>     > plans to
>     >     include it in my implementation afterwards. The Open Source
> middleware
>     >     vendor I'm working with was also extremely interested to see how
>     > Edgent can
>     >     be integrated into their platform.
>     >
>     >     I've never mailed to this mailing list or committed to this
> project
>     > before,
>     >     but I'm doing so now as I want to see Edgent grow into a mature
> Apache
>     >     project.
>     >     While it's difficult for me to delegate a lot of time at the
> moment, I
>     > will
>     >     contribute to the best of my ability in other aspects like
>     > documentation
>     >     and media.
>     >     Hope to get a response soon.
>     >
>     >     Thanks and best regards.
>     >
>     >     On Tue, Nov 6, 2018 at 7:18 AM Gayashan Amarasinghe <
>     >     gayashan.amarasinghe@gmail.com> wrote:
>     >
>     >     > Hi all,
>     >     >
>     >     > I am a phd student at University of Melbourne who is doing my
>     > research in
>     >     > Edge computing + Distributed Stream Processing. I have been
> following
>     >     > edgent (and I have followed Christofer to plc4x as well since
> I have
>     > seen
>     >     > him mentioning it here the work that he has done) for some
> time now.
>     > In
>     >     > fact I have sent couple of emails some time back but haven't
> been
>     > able to
>     >     > do anything significant. I have also used edgent in one of my
> ongoing
>     >     > research work even though it has not been published yet. I
> thought
>     > I'd send
>     >     > an email after seeing this thread.
>     >     >
>     >     > To simply put, I would like to help with the rebooting edgent,
> I
>     > think I
>     >     > would be able to use it in a mutually beneficial way such that
> I
>     > could use
>     >     > edgent to improve my research and get involved with edgent and
>     > edgent may
>     >     > get some features/bug fixes/etc as a result. But I have to be
> honest
>     > that
>     >     > being a phd student, I may not be able to be very active. But
> I am
>     > willing
>     >     > to do as much as possible in the free time I have. Let me know
> what
>     > you
>     >     > think and may be a way we can work together.
>     >     >
>     >     > Thank you.
>     >     >
>     >     > Best,
>     >     > Gayashan
>     >     >
>     >     >
>     >     > On Sun, Nov 4, 2018 at 9:09 PM Julian Feinauer <
>     >     > j.feinauer@pragmaticminds.de>
>     >     > wrote:
>     >     >
>     >     > > Hey Justin,
>     >     > >
>     >     > > thanks for your response.
>     >     > > Please don't take my words too literal (as I'm not (yet)
> used to
>     > many of
>     >     > > the ASF processes).
>     >     > > It's good to hear that you are still onboard. For me as
> rookie,
>     > it's good
>     >     > > to hear that we have multiple seasoned ASF member onboard.
>     >     > >
>     >     > > Regarding my statement with the attic.. what I really meant
> was to
>     >     > > "finish" the project officially, because currently, its kind
> of in
>     > a
>     >     > zombie
>     >     > > mode.
>     >     > >
>     >     > > But, I would definitely prefer a "revival" : )
>     >     > >
>     >     > > Julian
>     >     > >
>     >     > > Am 03.11.18, 22:59 schrieb "Justin Mclean" <
>     > justin@classsoftware.com>:
>     >     > >
>     >     > >     Hi,
>     >     > >
>     >     > >     > I think we (as we-that-are-on-the-list) have to decide
>     > whether we
>     >     > > want to (re-)start the project, building a new PPMC
> (currently we
>     > have
>     >     > > effectively one with chris) and start to do some work in
> edgent.
>     >     > >
>     >     > >     As far as PPMC (and mentor) duties I’m still around. The
> board
>     > likes
>     >     > a
>     >     > > minimum of 3 PPMC members, to graduate a minimal of 5 is
> seen as a
>     > good
>     >     > > number.
>     >     > >
>     >     > >     > Or whether we decide (or formally I guess PPMC and the
>     > mentors
>     >     > > decide) to move edgent to the attic.
>     >     > >
>     >     > >     Podlings that don’t make it, don’t end up in the attic,
> but
>     > the code
>     >     > > could continue to live on github.
>     >     > >
>     >     > >     Thanks,
>     >     > >     Justin
>     >     > >
>     >     > >
>     >     >
>     >
>     >
>     >     --
>     >
>     >     *Nuwan Jayawardene*
>     >     Computer Science Undergraduate at Informatics Institute of
>     > Technology(IIT),
>     >     Colombo
>     >     Tel - +94 71 181 3299
>     >     https://www.linkedin.com/in/nuwanjaya/
>     >     https://medium.com/@nuwanjaya
>     >
>     >
>     >
>
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message