Hi Chamikara, Chad
Thanks for your replies.
@Chamikara
I am already working on a Beam runner for Twister2, the runner is
functional for the most part even though I have not fully tested it. What I
was thinking was to find a way to transform just the sources using the Read
primitive that is used
I believe the flag was never relevant for PortableRunner. I might be wrong
as well. The flag affects a few bits in the core code and that is why the
solution cannot be by just setting the flag in Dataflow runner. It requires
some amount of clean up. I agree that it would be good to clean this up,
Thanks for offering to work on this! It would be awesome to have it. I can
say that we don't have that for Python ATM.
On Mon, Sep 16, 2019 at 10:56 AM Steve Niemitz wrote:
> Our experience has actually been that avro is more efficient than even
> parquet, but that might also be skewed from our
I believe that the Total shuffle data process counter counts the number of
bytes written to shuffle + the number of bytes read. So if you shuffle 1GB
of data, you should expect to see 2GB on the counter.
On Wed, Sep 18, 2019 at 2:39 PM Shannon Duncan
wrote:
> Ok just ran the job on a small
Sorry missed a part of the map output for flatten:
[image: image.png]
However the shuffle does show only 29.32 GB going into it but the output of
Total Shuffled data is 58.66 GB
[image: image.png]
On Wed, Sep 18, 2019 at 4:39 PM Shannon Duncan
wrote:
> Ok just ran the job on a small input
Ok just ran the job on a small input and did not specify numShards. so it's
literally just:
.apply("WriteLines", TextIO.write().to(options.getOutput()));
Output of map for join:
[image: image.png]
Details of Shuffle:
[image: image.png]
Reported Bytes Shuffled:
[image: image.png]
On Wed, Sep
Hi Pulasthi,
Just to mirror what Cham said, it would be a non-starter to try to use a
Beam IO source in another framework: to make them work, you'd have to build
something that executes them with their expected protocol, and that would
look an awful lot like a Beam runner. It makes more sense to
On Wed, Sep 18, 2019 at 2:12 PM Shannon Duncan
wrote:
> I will attempt to do without sharding (though I believe we did do a run
> without shards and it incurred the extra shuffle costs).
>
It shouldn't. There will be a shuffle, but that shuffle should contain a
small amount of data (essentially
Hi Pulasthi,
This might be possible but I don't know if anybody has done this. API of
Beam sources are no different from other Beam PTransforms and we highly
recommend hiding away various implementations of source framework related
abstractions in a composite transform [1]. So what you are
I will attempt to do without sharding (though I believe we did do a run
without shards and it incurred the extra shuffle costs).
Pipeline is simple.
The only shuffle that is explicitly defined is the shuffle after merging
files together into a single PCollection (Flatten Transform).
So it's a
In that case you should be able to leave sharding unspecified, and you
won't incur the extra shuffle. Specifying explicit sharding is generally
necessary only for streaming.
On Wed, Sep 18, 2019 at 2:06 PM Shannon Duncan
wrote:
> batch on dataflowRunner.
>
> On Wed, Sep 18, 2019 at 4:05 PM
batch on dataflowRunner.
On Wed, Sep 18, 2019 at 4:05 PM Reuven Lax wrote:
> Are you using streaming or batch? Also which runner are you using?
>
> On Wed, Sep 18, 2019 at 1:57 PM Shannon Duncan
> wrote:
>
>> So I followed up on why TextIO shuffles and dug into the code some. It is
>> using
Are you specifying the number of shards to write to:
https://github.com/apache/beam/blob/master/sdks/java/core/src/main/java/org/apache/beam/sdk/io/TextIO.java#L859
If so, this will incur an additional shuffle to re-distribute data written
by all workers into the given number of shards before
Are you using streaming or batch? Also which runner are you using?
On Wed, Sep 18, 2019 at 1:57 PM Shannon Duncan
wrote:
> So I followed up on why TextIO shuffles and dug into the code some. It is
> using the shards and getting all the values into a keyed group to write to
> a single file.
>
>
What you propose with a writer per bundle is definitely possible, but I
expect the blocker is that in most cases the runner has control of bundle
sizes and there's nothing exposed to the user to control that. I've wanted
to do similar, but found average bundle sizes in my case on Dataflow to be
so
So I followed up on why TextIO shuffles and dug into the code some. It is
using the shards and getting all the values into a keyed group to write to
a single file.
However... I wonder if there is way to just take the records that are on a
worker and write them out. Thus not needing a shard number
Hi Dev's
We have a big data processing framework named Twister2, and wanted to know
if there is any way we could leverage the I/O Transforms that are built
into Apache Beam externally. That is rather than using it in a Beam
pipeline just use them as data sources in our project. Just wanted to
Hi Rahul,
The Beam Summit committee is working on this at the moment. Stay tuned.
Thanks,
Max
On 18.09.19 11:39, rahul patwari wrote:
Hi,
The videos of Beam Summit that has happened recently have disappeared
from YouTube Apache Beam channel.
Is uploading the videos a WIP?
Thanks,
Rahul
I disagree that this flag is obsolete. It is still serving a purpose for batch
users using dataflow runner and that is decent chunk of beam python users.
It is obsolete for the PortableRunner. If the Dataflow Runner needs this
flag, couldn't we simply add it there? As far as I know Dataflow
Hi,
I would love to join the call.
Can you also share the meeting invitation with me?
Thanks,
Rahul
On Wed 18 Sep, 2019, 11:48 PM Xinyu Liu, wrote:
> Alexey and Etienne: I'm very happy to join the sync-up meeting. Please
> forward the meeting info to me. I am based in California, US and
Hi,
The videos of Beam Summit that has happened recently have disappeared from
YouTube Apache Beam channel.
Is uploading the videos a WIP?
Thanks,
Rahul
Alexey and Etienne: I'm very happy to join the sync-up meeting. Please
forward the meeting info to me. I am based in California, US and hopefully
the time will work :).
Thanks,
Xinyu
On Wed, Sep 18, 2019 at 6:39 AM Etienne Chauchot
wrote:
> Hi Xinyu,
>
> Thanks for offering help ! My comments
Hi Xinyu,
Thanks for offering help ! My comments are inline:
Le vendredi 13 septembre 2019 à 12:16 -0700, Xinyu Liu a écrit :
> Hi, Etienne,
> The slides are very informative! Thanks for sharing the details about how the
> Beam API are mapped into Spark
> Structural Streaming.
Thanks !
> We
Hi Rui,Thanks for proposing to contribute to this new runner !
Here are the pointers:- SS runner branch:
https://github.com/apache/beam/tree/spark-runner_structured-streaming- spark
design doc for multiple watermarks support:
24 matches
Mail list logo