Of Data Dropboxes and Data Gloveboxes

Explore the intriguing parallels between Cold War era nuclear isolation techniques and modern data science security in this thought-provoking conference talk. Discover how the concept of nuclear gloveboxes can be applied to protect sensitive data in today's digital landscape. Learn about innovative approaches like Data Dropboxes and Data Gloveboxes, designed to limit data exfiltration and ensure secure, regularized data production. Delve into the philosophy and implementation of these security measures, including blind data input with checksum verification and controlled data manipulation processes. Gain insights into topics such as Hadoop cluster security, HDFS WebHDFS API filtering, isolation techniques, and the importance of firewalls in maintaining data integrity. Understand how these concepts can be applied to create secure environments for data scientists working with highly sensitive information, mirroring the precautions taken by nuclear engineers during the Cold War.

Intro
Bloomberg By the Numbers
Handling Material
Properties of Radioactive Materials
Properties of Data
Nuclear Material (non) Proliferation
Data Material Proliferation
Clean-Up Is Messy
Hadoop Cluster
Lock Everything Down
DropboxFilter for HDFS WebHDFS API
DropboxFilter for HDFS (Architecture)
DropboxFilter for HDFS (Examples)
Plutonium Enclave
Data Enclave
Isolation Glovebox
Glovebox Production Line
Data Glovebox (Leaded Pane of Glass)
Data Glovebox (Glove Ports & Robotics)
Data Glovebox (Pass-Through)
Firewalls (Workload and User Isolation)
Take out the Trash
Private Temporary Directories
Keeping the Pipes Flowing
YARN Network Isolation (Example)
Firewalls Are Important