The Future Architecture of a Data Lake: In-memory Data Exchange Platform Using Tachyon and Apache Spark

October 14, 2014 Paul M. Davis

featured-tachyon-sparkThe next generation of data lake technology will leverage the availability of in-memory processing, with an architecture that supports multiple data analytics workloads within a single environment. Pivotal has led the way with the Pivotal Big Data Suite, working to realize a data lake that serves as a central repository for all the data within an enterprise. This model brings true SQL-level processing to all data, whether disk-based or in-memory, and enables polyglot persistence, allowing multiple computation paradigms to process data in-place.

However, memory-centric platforms for computation and analytics are becoming increasingly popular, especially as the cost per GB of high performance RAM continues to drop. The performance gap between in-memory computation and on-disk analysis is growing larger, and companies are adding memory-centric databases such as Pivotal Gemfire to their deployments at an increasing rate.

Pivotal is revolutionizing the data lake with an architecture that builds upon disk-based storage with memory-centric processing frameworks. In partnership with the AMPLab at UC Berkeley, Pivotal envisions this future architecture will incorporate an in-memory data exchange platform based on Tachyon and in-memory compute layer augmented by Apache Spark.

The result is a next-generation date lake implementation based on Spark and Tachyon, which Pivotal is referring to as a “butterfly architecture.” Within this model, Tachyon provides an efficient memory-centric caching layer for disparate data sources, and allows the tracking of data lineage, independent of the computation framework. It will serve as an efficient memory-based data exchange layer within the data lake, and is pluggable, enabling existing storage and processing systems to co-exist with the new framework.

Tachyon is a memory-centric, fault-tolerant distributed file system which enables reliable data exchange at in-memory speed across cluster frameworks. Pivotal envisions Tachyon as a central data exchange layer for the Pivotal Big Data Suite. To this end, Pivotal is actively dedicating resources to Tachyon’s development and stands as the number one corporate contributor to the Tachyon code base.

The Tachyon project is spearheaded by Haoyuan Li, a PhD candidate at the UC Berkeley AMPLab. Standing as the fastest-growing project in all of AMPLab history, surpassing Mesos and even Spark itself, Tachyon is currently a GitHub project developed under the Apache License.

In response to the rapidly growing interest in Tachyon from a number of industry players, Pivotal is leading the charge on helping the project formalize and tighten its governance model, thus allowing even faster rate of innovation with a more predictable roadmap. The journey is expected to culminate with Tachyon entering Apache Software Foundation Incubator program similar to the path of other big data projects that came from AMPLab.

In addition to contributing to Tachyon’s development and integration with its Big Data Suite, Pivotal is supporting the project through Research Fellowship Program. The program will advance the Tacyhon project through fellowship and internship support, and will contribute ongoing development resources, as well as benchmarks and use cases.

Pivotal believes that Tachyon will revolutionize how in-memory processing works with file storage, such as HDFS. Pivotal partner EMC is already looking into integration of Tachyon with the advanced flash storage product DSSD, as well as Isilon technologies. Expressing this enthusiasm, EMC’s Chief Technology Officer John Roese stated, ”EMC is excited to join with Pivotal in supporting and fostering the rapidly innovating Big Data ecosystem. The next wave of real-time analytics will be made possible by technologies such as Apache Spark and Tachyon, in combination with innovative storage products such as EMC’s recent DSSD acquisition.”

New York Meetup
Title: Evolution of Data Architectures: Pivotal’s Data Lake Vision for 2015
Where: 625 Avenue of Americas, 2nd Floor, New York, NY
When: Weds, October 15th, 7 PM
Pivotal New York

To learn more about Tachyon, attend Pivotal’s New York meetup, “Evolution of Data Architectures: Pivotal’s Data Lake Vision for 2015,” on Wednesday October 15th at 7pm at the Pivotal Labs New York Office, located on 625 Avenue of Americas, 2nd Floor, New York, NY. For those unable to attend, the talk will be recorded and available online following the event.

About the Author


Boston Pizza Mobile App Demo
Boston Pizza Mobile App Demo

Boston Pizza recently released its mobile ordering app for iOS in conjunction with the launch of a new digi...

What to do with a bullet-pointed list of features
What to do with a bullet-pointed list of features

Our client came in with a short bullet-pointed list of features they want for their new iPhone app. Part of...

SpringOne 2022

Register Now