Parallel Data Laboratory Talk
- Remote Access - Zoom
- Virtual Presentation - ET
- PAT HELLAND
- Principal Architect
Autonomous Computing: Working across Trust Boundaries
- 12:00 pm — Welcomes and Introductions → Greg Ganger, Director, Parallel Data Laboratory
- ~12:10 pm — Guest Speaker → Pat Helland
Businesses have accomplished work across trust boundaries for centuries. When I was a kid, this was done using specific patterns of paper forms that physically moved across companies and across departments within companies. Similarly, data was used in patterns that supported this work. All of this happened without database transactions spanning these distrusting environments and it still does.
This talk discusses this pattern by introducing three notions with whimsical names. A "fiefdom" refers to computation and data that doesn't trust outsiders. It accepts messages and performs work to accomplish business with outsiders but doesn't trust them and would never do a database transaction locking up records waiting for the outsider. An "emissary" is code that lives outside the trust boundary but has been designed to make it easy to work with the fiefdom, its messaging patterns, and its expected data. Finally, a "collaboration" is a set of related messages that flow in and out of the trust boundary to accomplish business.
We will discuss how these patterns were used before computers and how they are still key to our lives today. Internal to large business operations today, these patterns are essential to scaling, replication, long-running solutions, and more. Autonomy is key to our complex systems.
Pat Helland has been building databases, transaction systems, messaging systems, application platforms, and storage systems since 1978 at a number of companies including Tandem, Microsoft, and Amazon. He has a regular column in ACM Queue called: "Escaping the Singularity: It's Not Your Grandmother's DBMS Anymore" and blogs at Scattered Thoughts on Distributed Systems. Pat has worked at Salesforce since 2012 developing exciting data infrastructure for the cloud. You can follow him on Twitter at @pathelland.