China Unicom is one of the five largest telecom operators in the world. China Unicom’s booming business in 4G and 5G networks has to serve an exploding base of hundreds of millions of smartphone users. This unprecedented growth brought enormous challenges and new requirements to the data processing infrastructure. The previous generation of its data processing system was based on IBM midrange computers, Oracle databases, and EMC storage devices. This architecture could not scale to process the amounts of data generated by the rapidly expanding number of mobile users. Even after deploying Hadoop and Greenplum database, it was still difficult to cover critical business scenarios with their varying massive data processing requirements.
Tag: apache spark
Tachyon is a memory-centric fault-tolerant distributed storage system, which enables reliable file sharing at memory-speed. It originated from AMPLab, UC Berkeley in 2012, the same lab produced Apache Mesos and Apache Spark. Soon later, it became an open source project and is deployed at many companies. Since then, Tachyon has attracted more than 200 contributors from over 50 institutions. In 2015, company Tachyon Nexus was founded to further accelerate the development of Tachyon. In this talk, we will review Tachyon’s new features, deployments, and developments in 2015, and look into 2016.
During the past several years, Spark has significantly changed the landscape of big data computing. It improves performance of various applications dramatically. However, in certain Spark use cases, the bottleneck is in the I/O stack. In this talk, we will introduce Tachyon, a distributed memory-centric storage system. In addition, we will talk about several production use cases where Tachyon further improves Spark applications’ performance by orders of magnitude.
A few months ago, Baidu deployed Alluxio to accelerate its big data analytics workload. Bin Fan and Haojun Wang explain why Baidu chose Alluxio, as well as the details of how they achieved a 30x speedup with Alluxio in their production environment with hundreds of machines. Based on the success of the big data analytics engine, Baidu is currently expanding the Alluxio and Spark infrastructure to accelerate other applications, such as machine learning.
In this talk, Haoyuan Li, co-creator of Tachyon (and a founding committer of Spark) and CEO of Tachyon Nexus will explain how the next wave of innovation in storage will be driven by separating the functional layer from the persistent storage layer, and how memory-centric architecture through Tachyon is making this possible. Li will describe the future of distributed file storage and highlight how Tachyon supports specific use cases.
Throughout our four-year history, Scala and Scale By the Bay is leading the way on evangelizing and understansing modern software architectures. We have the best set of them here, including Akka, Kafka, Spark, Finagle, Lagom, and so on. How do they come together in a SMACK / MIND Stack? What are the best practices to follow and pitfalls to avoid? This panels of experienced practitioners will discuss and illuminate it all.
Using intermediate APIs means developers can learn just one framework and still access features offered by different technologies. It means writing job logic only once and being able to test it easily on a new underlying service with no effort. Not only is modularity a win for users but it means creators of execution frameworks and storage systems can focus on performance and capability without having to worry about API maintenance.
In this talk, we briefly introduce Alluxio, present several ways how Alluxio can help Spark be more effective, show benchmark results with Spark RDDs and DataFrames, and describe production deployments both Alluxio and Spark working together. In the meantime, we will provide live demos for some of the use cases.
Using Alluxio to Improve Spark & Hadoop HDFS System Performance and Reliability [Chinese]