spark-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Felix Cheung <>
Subject Re: Revisiting Online serving of Spark models?
Date Wed, 30 May 2018 16:32:27 GMT

Thank you! Let’s meet then

June 6 4pm

Moscone West Convention Center
800 Howard Street, San Francisco, CA 94103

Ground floor (outside of conference area - should be available for all) - we will meet and
decide where to go

(Would not send invite because that would be too much noise for dev@)

To paraphrase Joseph, we will use this to kick off the discusssion and post notes after and
follow up online. As for Seattle, I would be very interested to meet in person lateen and
discuss ;)

From: Saikat Kanjilal <>
Sent: Tuesday, May 29, 2018 11:46 AM
Subject: Re: Revisiting Online serving of Spark models?
To: Maximiliano Felice <>
Cc: Felix Cheung <>, Holden Karau <>,
Joseph Bradley <>, Leif Walsh <>, dev

Would love to join but am in Seattle, thoughts on how to make this work?


Sent from my iPhone

On May 29, 2018, at 10:35 AM, Maximiliano Felice <<>>

Big +1 to a meeting with fresh air.

Could anyone send the invites? I don't really know which is the place Holden is talking about.

2018-05-29 14:27 GMT-03:00 Felix Cheung <<>>:
You had me at blue bottle!

From: Holden Karau <<>>
Sent: Tuesday, May 29, 2018 9:47 AM
Subject: Re: Revisiting Online serving of Spark models?
To: Felix Cheung <<>>
Cc: Saikat Kanjilal <<>>, Maximiliano
Felice <<>>, Joseph
Bradley <<>>, Leif Walsh <<>>,
dev <<>>

I'm down for that, we could all go for a walk maybe to the mint plazaa blue bottle and grab
coffee (if the weather holds have our design meeting outside :p)?

On Tue, May 29, 2018 at 9:37 AM, Felix Cheung <<>>

From: Felix Cheung <<>>
Sent: Saturday, May 26, 2018 1:05:29 PM
To: Saikat Kanjilal; Maximiliano Felice; Joseph Bradley
Cc: Leif Walsh; Holden Karau; dev

Subject: Re: Revisiting Online serving of Spark models?

Hi! How about we meet the community and discuss on June 6 4pm at (near) the Summit?

(I propose we meet at the venue entrance so we could accommodate people might not be in the

From: Saikat Kanjilal <<>>
Sent: Tuesday, May 22, 2018 7:47:07 AM
To: Maximiliano Felice
Cc: Leif Walsh; Felix Cheung; Holden Karau; Joseph Bradley; dev
Subject: Re: Revisiting Online serving of Spark models?

I’m in the same exact boat as Maximiliano and have use cases as well for model serving and
would love to join this discussion.

Sent from my iPhone

On May 22, 2018, at 6:39 AM, Maximiliano Felice <<>>


I'm don't usually write a lot on this list but I keep up to date with the discussions and
I'm a heavy user of Spark. This topic caught my attention, as we're currently facing this
issue at work. I'm attending to the summit and was wondering if it would it be possible for
me to join that meeting. I might be able to share some helpful usecases and ideas.

Maximiliano Felice

El mar., 22 de may. de 2018 9:14 AM, Leif Walsh <<>>
I’m with you on json being more readable than parquet, but we’ve had success using pyarrow’s
parquet reader and have been quite happy with it so far. If your target is python (and probably
if not now, then soon, R), you should look in to it.

On Mon, May 21, 2018 at 16:52 Joseph Bradley <<>>
Regarding model reading and writing, I'll give quick thoughts here:
* Our approach was to use the same format but write JSON instead of Parquet.  It's easier
to parse JSON without Spark, and using the same format simplifies architecture.  Plus, some
people want to check files into version control, and JSON is nice for that.
* The reader/writer APIs could be extended to take format parameters (just like DataFrame
reader/writers) to handle JSON (and maybe, eventually, handle Parquet in the online serving

This would be a big project, so proposing a SPIP might be best.  If people are around at the
Spark Summit, that could be a good time to meet up & then post notes back to the dev list.

On Sun, May 20, 2018 at 8:11 PM, Felix Cheung <<>>
Specifically I’d like bring part of the discussion to Model and PipelineModel, and various
ModelReader and SharedReadWrite implementations that rely on SparkContext. This is a big blocker
on reusing  trained models outside of Spark for online serving.

What’s the next step? Would folks be interested in getting together to discuss/get some

From: Felix Cheung <<>>
Sent: Thursday, May 10, 2018 10:10 AM
Subject: Re: Revisiting Online serving of Spark models?
To: Holden Karau <<>>, Joseph Bradley
Cc: dev <<>>

Huge +1 on this!

________________________________<> <<>>
on behalf of Holden Karau <<>>
Sent: Thursday, May 10, 2018 9:39:26 AM
To: Joseph Bradley
Cc: dev
Subject: Re: Revisiting Online serving of Spark models?

On Thu, May 10, 2018 at 9:25 AM, Joseph Bradley <<>>
Thanks for bringing this up Holden!  I'm a strong supporter of this.

Awesome! I'm glad other folks think something like this belongs in Spark.
This was one of the original goals for mllib-local: to have local versions of MLlib models
which could be deployed without the big Spark JARs and without a SparkContext or SparkSession.
 There are related commercial offerings like this : ) but the overhead of maintaining those
offerings is pretty high.  Building good APIs within MLlib to avoid copying logic across libraries
will be well worth it.

We've talked about this need at Databricks and have also been syncing with the creators of
MLeap.  It'd be great to get this functionality into Spark itself.  Some thoughts:
* It'd be valuable to have this go beyond adding transform() methods taking a Row to the current
Models.  Instead, it would be ideal to have local, lightweight versions of models in mllib-local,
outside of the main mllib package (for easier deployment with smaller & fewer dependencies).
* Supporting Pipelines is important.  For this, it would be ideal to utilize elements of Spark
SQL, particularly Rows and Types, which could be moved into a local sql package.
* This architecture may require some awkward APIs currently to have model prediction logic
in mllib-local, local model classes in mllib-local, and regular (DataFrame-friendly) model
classes in mllib.  We might find it helpful to break some DeveloperApis in Spark 3.0 to facilitate
this architecture while making it feasible for 3rd party developers to extend MLlib APIs (especially
in Java).
I agree this could be interesting, and feed into the other discussion around when (or if)
we should be considering Spark 3.0
I _think_ we could probably do it with optional traits people could mix in to avoid breaking
the current APIs but I could be wrong on that point.
* It could also be worth discussing local DataFrames.  They might not be as important as per-Row
transformations, but they would be helpful for batching for higher throughput.
That could be interesting as well.

I'll be interested to hear others' thoughts too!


On Wed, May 9, 2018 at 7:18 AM, Holden Karau <<>>
Hi y'all,

With the renewed interest in ML in Apache Spark now seems like a good a time as any to revisit
the online serving situation in Spark ML. DB & other's have done some excellent working
moving a lot of the necessary tools into a local linear algebra package that doesn't depend
on having a SparkContext.

There are a few different commercial and non-commercial solutions round this, but currently
our individual transform/predict methods are private so they either need to copy or re-implement
(or put them selves in org.apache.spark) to access them. How would folks feel about adding
a new trait for ML pipeline stages to expose to do transformation of single element inputs
(or local collections) that could be optionally implemented by stages which support this?
That way we can have less copy and paste code possibly getting out of sync with our model

I think continuing to have on-line serving grow in different projects is probably the right
path, forward (folks have different needs), but I'd love to see us make it simpler for other
projects to build reliable serving tools.

I realize this maybe puts some of the folks in an awkward position with their own commercial
offerings, but hopefully if we make it easier for everyone the commercial vendors can benefit
as well.


Holden :)



Joseph Bradley

Software Engineer - Machine Learning

Databricks, Inc.




Joseph Bradley

Software Engineer - Machine Learning

Databricks, Inc.




View raw message