flink-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "wenlong.lwl" <wenlong88....@gmail.com>
Subject Re: [DISCUSS] FLIP-17 Side Inputs
Date Tue, 07 Mar 2017 13:39:48 GMT
Hi Aljoscha, thank you for the proposal, it is great to hear about the
progress in side input.

Following is my point of view:
1. I think there may be an option to block the processing of the main input
instead of buffer the data in state because in production, the through put
of the main input is usually much larger, and buffering the data before the
side input may slow down the preparing of side input since the i-o and
computing resources are always limited.
2. another issue may need to be disscussed: how can we do checkpointing
with side input, because static side input may finish soon once started
which will stop the checkpointing.
3. I agree with Gyula that user should be able to determines when a side
input is ready? Maybe we can do it one step further: whether users can
determine a operator with multiple inputs to process which input each time
or not?  It would be more flexible.


Best Regards!
Wenlong

On 7 March 2017 at 18:39, Ventura Del Monte <venturadelmonte@gmail.com>
wrote:

> Hi Aljoscha,
>
> Thank you for the proposal and for bringing up again this discussion.
>
> Regarding the implementation aspect,I would say the first way could
> be easier/faster to implement but it could add some overhead when
> dealing with multiple side inputs through the current 2-streams union
> transform. I tried the second option myself as it has less overhead
> but then the outcome was something close to a N-ary operator consuming
> first each side input while buffering the main one.
> Therefore, I would choose the third option as it is more generic
> and might help also in other scenarios, although its implementation
> requires more effort.
> I also agree with Gyula, I think the user should be allowed to define the
> condition that determines when a side input is ready, e.g., load the side
> input first, incrementally update the side input.
>
> Best,
> Ventura
>
>
>
>
>
>
> This message, for the D. Lgs n. 196/2003 (Privacy Code), may contain
> confidential and/or privileged information. If you are not the addressee or
> authorized to receive this for the addressee, you must not use, copy,
> disclose or take any action based on this message or any information
> herein. If you have received this message in error, please advise the
> sender immediately by reply e-mail and delete this message. Thank you for
> your cooperation.
>
> On Mon, Mar 6, 2017 at 3:50 PM, Gyula Fóra <gyula.fora@gmail.com> wrote:
>
> > Hi Aljoscha,
> >
> > Thank you for the nice proposal!
> >
> > I think it would make sense to allow user's to affect the readiness of
> the
> > side input. I think making it ready when the first element arrives is
> only
> > slightly better then making it always ready from usability perspective.
> For
> > instance if I am joining against a static data set I want to wait for the
> > whole set before making it ready. This could be exposed as a user defined
> > condition that could also recognize bounded inputs maybe.
> >
> > Maybe we could also add an aggregating (merging) side input type, that
> > could work as a broadcast state.
> >
> > What do you think?
> >
> > Gyula
> >
> > Aljoscha Krettek <aljoscha@apache.org> ezt írta (időpont: 2017. márc.
> 6.,
> > H, 15:18):
> >
> > > Hi Folks,
> > >
> > > I would like to finally agree on a plan for implementing side inputs in
> > > Flink. There has already been an attempt to come to consensus [1],
> which
> > > resulted in two design documents. I tried to consolidate those two and
> > > also added a section about implementation plans. This is the resulting
> > > FLIP:
> > >
> > > https://cwiki.apache.org/confluence/display/FLINK/FLIP-
> > 17+Side+Inputs+for+DataStream+API
> > >
> > >
> > > In terms of semantics I tried to go with the minimal viable solution.
> > > The part that needs discussing is how we want to implement this. I
> > > outlined three possible implementation plans in the FLIP but what it
> > > boils down to is that we need to introduce some way of getting several
> > > inputs into an operator/task.
> > >
> > >
> > > Please have a look at the doc and let us know what you think.
> > >
> > >
> > >
> > > Best,
> > >
> > > Aljoscha
> > >
> > >
> > >
> > > [1]
> > > https://lists.apache.org/thread.html/797df0ba066151b77c7951fd7d603a
> > 8afd7023920d0607a0c6337db3@1462181294@%3Cdev.flink.apache.org%3E
> > >
> >
>

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