Hi Dmitry!

The streaming runtime makes a conscious decision to not merge streams as in
an ordered merge.
The reason is that this is at large scale typically bad for scalability /
network performance.
Also, in certain DAGs, it may lead to deadlocks.

Even the two input operator delivers records on a low level in a
first-come-first-serve order as driven by network events (NIO events).

Flink's operators tolerate out-of-order records to compensate for that.
Overall, that seemed the more scalable design to us.
Can your use case follow a similar approach?

Stephan



On Tue, Jan 17, 2017 at 10:57 AM, Dmitry Golubets <dgolub...@gmail.com>
wrote:

> Hi Timo,
>
> I don't have any key to join on, so I'm not sure Window Join would work
> for me.
>
> Can I implement my own "low level" operator in any way?
> I would appreciate if you can give me a hint or a link to example of how
> to do it.
>
>
>
> Best regards,
> Dmitry
>
> On Tue, Jan 17, 2017 at 9:24 AM, Timo Walther <twal...@apache.org> wrote:
>
>> Hi Dmitry,
>>
>> the runtime supports an arbitrary number of inputs, however, the API does
>> currently not provide a convenient way. You could use the "union" operator
>> to reduce the number of inputs. Otherwise I think you have to implement
>> your own operator. That depends on your use case though.
>>
>> You can maintain backpressure by using Flink's operator state. But did
>> you also thought about a Window Join instead?
>>
>> I hope that helps.
>>
>> Timo
>>
>>
>>
>>
>> Am 17/01/17 um 00:20 schrieb Dmitry Golubets:
>>
>> Hi,
>>
>> there are only *two *interfaces defined at the moment:
>> *OneInputStreamOperator*
>> and
>> *TwoInputStreamOperator.*
>>
>> Is there any way to define an operator with arbitrary number of inputs?
>>
>> My another concern is how to maintain *backpressure *in the operator?
>> Let's say I read events from two Kafka sources, both of which are ordered
>> by time. I want to merge them keeping the global order. But to do it, I
>> need to stop block one input if another one has no data yet.
>>
>> Best regards,
>> Dmitry
>>
>>
>>
>

Reply via email to