June 4th, 2020

Sparkling Water 3.30.0.3 is out

RSS icon RSS Category: H2O, Open Source, Sparkling Water

Sparkling Water is about making machine learning simple, speedy, and scalable with Apache Spark. This blog provides an overview of the following new features: 

  • No H2O Client on Spark Driver
  • Speedups
  • Automatic String conversion to Categoricals

No H2O Client on Spark Driver

Previously, Sparkling Water always started worker nodes either on Spark executors in case of internal or external backend. To be able to communicate with such created worker nodes, we needed to start another instance of H2O on the Spark driver. This approach, however, has several disadvantages. 

In order to overcome the disadvantages from our previous approach, we decided to completely replace the H2O client with a lightweight REST API client which runs on the Spark driver. Our new approach significantly reduced the design limits of Sparkling Water. One of the major changes was removing the H2O client from the Spark driver in both PySparkling and RSparkling backendsThis new approach has several benefits:

Easier to Deploy and Manage

  • Because the H2O client was just standard H2O with a special flag, it required two ports to be able to communicate with the rest of the cluster. One for the API and one internal port. Currently, we no longer need the internal port and all the communication goes via the API port. We also support SSL via HTTPS.
  • In the case of an external backend, it is now possible to reconnect to a new pre-existing H2O cluster during run-time. This was not possible before as the H2O client was tied to a single H2O cluster.
  • We avoid issues with connections when the H2O client & the rest of the H2O worker nodes run on different networks. Previously we had to provide extra configurations such as client IP or client network address in order for the client to be able to connect. These requirements are now lifted and there is no need for such configuration.
  • Since the H2O client was in some terms a regular H2O node, Xgboost was performing some computations on the Spark driver which is not desired. This is no longer the case with the new version.

More Secure

  • In some environments, the H2O client can reveal security issues. By not using the client and using single, for example, secured HTTP connection, we ensure that the cluster and the communication are secured.

Enable Sparkling Water for Kubernetes

  • This client-less solution simplifies Kubernetes deployments and enables Sparkling Water to run in cluster mode in Kubernetes.

Speedups

The version of 3.30.0.3 also includes several speedups when it comes to conversion time from H2O to Spark frame.  The following examples show speedups on several datasets by 50%.

Benchmark

Sparkling Water Versions
Dataset pre-3.30.0.3 

(milliseconds)

3.30.0.3
(milliseconds)
Columns Rows Speedup
airlines-1m 5031 1603 9 1m Almost 5x
airlines-10m 7223 7151 9 10m Same
springleaf 24463 13951 1936 101810 Almost 2x

We can see that this speedup affects smaller datasets as we improve the internal protocol and ovoid creation of some unnecessary objects.

This speed comparison in data frame conversion was done using the H2OFrameToDataFrameConversionBenchmark in our benchmark. You can try it yourself or, better yet, try it with your data and let us know what performance you are seeing. 

Simplifying String Conversion to Categoricals

Spark to H2O conversion on H2OContext now automatically converts strings to categorical values. This ensures that the frame created during the hc.asH2OFrame call has the same properties as the frame imported directly using h2o.import_file. We hope that his change will again simplify the Sparkling Water usage.

For example, in PySparkling API, the previous code was:

frame = hc.asH2OFrame(sparkDataFrame)
frame[“str_col”] = frame[“str_col”].asfactor()

In the new version of the API, the conversion to categorical is done automatically, so the code you need to run is only:

frame = hc.asH2OFrame(sparkDataFrame)

Upcoming Changes

Please note that the benefits of REST API clients are available only in RSparkling and PySparkling so far. We are actively working on making sure the Scala implementation can run without the client as well and we are getting really close to achieving that.

Several other changes we are planning to incorporate to Sparkling Water soon:

  • Kubernetes support for Sparkling Water.
  • Expose remaining algorithms and stack ensembles to Sparkling Water API

Stay tuned and we hope you enjoy the new features in Sparkling Water.

Credits

This new Sparkling Water release is brought to you by Jakub Háva and Marek Novotný.

About the Author

Jakub Hava
Jakub Hava

Jakub (or “Kuba” as we call him) completed his Bachelor’s Degree in Computer Science and Master’s Degree in Software Systems at Charles University in Prague. As a bachelor’s thesis, Kuba wrote a small platform for distributed computing of any types of tasks. During his master’s degree studies, he developed a cluster monitoring tool for JVM based languages which makes debugging and reasoning the performance of distributed systems easier using a concept called distributed stack traces. Kuba enjoys dealing with problems and learning new programming languages. At H2O.ai, Kuba works on Sparkling Water. Aside from programming, Kuba enjoys exploring new cultures and bouldering. He’s also a big fan of tea preparation and the associated ceremony.

Leave a Reply

A Inteligência Artificial está transformando e alavancando negócios. Entenda como e por quê

Você sabia que inteligência artificial e machine learning não são conceitos novos? Pois eles surgiram

June 26, 2020 - by Daniel Garbuglio
On-Ramp to AI

The path to democratize AI starts with one class Artificial Intelligence (AI) is like a superhighway,

June 11, 2020 - by Rafael Coss
From GLM to GBM – Part 1

How an Economics Nobel Prize could revolutionize insurance and lending Part 1: A New Solution to

June 9, 2020 - by Michael Proksch and Patrick Hall
Fallback Featured Image
Machine Learning in the Age of AutoML

“Just as electricity transformed almost everything 100 years ago, today I actually have a hard

May 15, 2020 - by Parul Pandey
Are All Your AI and ML Models Wrong?

We are living in unprecedented times. Our society and economy are experiencing shocks beyond anything

May 5, 2020 - by James Orton
Lessons of COVID-19 and Moving Forward: Key Takeaways

This week, we hosted our second virtual panel focused on how AI can empower healthcare

May 1, 2020 - by Ingrid Burton

Join the AI Revolution

Subscribe, read the documentation, download or contact us.

Subscribe to the Newsletter

Start Your 21-Day Free Trial Today

Get It Now
Desktop img