Real World Use Cases for Tachyon - A Memory-Centric Distributed Storage System

Explore the real-world applications of Tachyon, a memory-centric distributed storage system, in this 40-minute Devoxx conference talk. Discover how Tachyon addresses the storage bottleneck in big data processing by enabling reliable file sharing at memory-speed across cluster frameworks like Apache Spark, MapReduce, and Apache Flink. Learn about Tachyon's architecture, key features, and performance benefits, including its Hadoop compatibility and fault-tolerance capabilities. Examine production use cases, such as the Sauce Company implementation, and understand how Tachyon integrates with existing big data ecosystems. Delve into new features, tiered storage concepts, and deployment strategies. Gain insights into Tachyon's growing community, with over 170 contributors from major tech companies, and its role in the Berkeley Data Analytics Stack and Fedora distribution.

Intro
About Tachyon
More than 170 contributors
Fedora distribution
Comparison
Community
Contributors Users
Stack
Performance
Price Trend
Spark
Hana
How does Tachyon work
Two features
Architecture
Lineage
Spark example
Spark issues
Tachyon solution
Memory in Spark
Memory in Tag
Application duplication
Solution
Production Use Case
Sauce Company Use Case
New Features
Benefits of Tachyon
Tiered Storage Concept
Storage Tier Configuration
Policy
Transparent Naming
Unified Naming
Remote Right
Tachyon Deployment
Tachyon Security
Collaboration
Questions