Explore a conference talk detailing Data Domain's cloud tier capability for its on-premises storage appliance, focusing on key innovations that support essential customer interactions. Learn about architectural changes and new algorithms addressing the impact of deduplicated storage and cloud object storage challenges. Discover how customers estimate space savings, transfer files to and from the cloud, and manage file deletion in the cloud tier. Gain insights into real-world deployments, including examples of customers moving over 20PB of logical data to the cloud tier and achieving compression factors of 40× or more, resulting in significant cost savings. Delve into topics such as file representation, active and cloud tier architecture, deduplication system challenges, physical scanning, efficient data seeding and incremental transfers, and dead data cleaning in the cloud tier. Examine the evaluation methodology, field analysis, and summary of results from this USENIX ATC '19 presentation.
Data Domain Cloud Tier - Backup Here, Backup There, Deduplicated Everywhere!
Overview
Syllabus
Intro
Data Domain Appliance
Data Domain Cloud Tier
File Representation in Data Domain Filesystem
Active Tier and Cloud Tier Architecture
Challenges in Deduplication System
Requirements for Cloud Tier
Physical Scanning
How much space will be freed if we move data to cloud tier?
How can we seed data efficiently to the cloud tier?
How can we move incremental data to cloud tier?
How can we efficiently clean dead data on cloud tier?
Evaluation Methodology
Field Analysis
Summary of Field Evaluation Results
Internal Evaluation !
Conclusion
THE ADVANCED COMPUTING SYSTEMS ASSOCIATION
Taught by
USENIX
