February Distributed Systems Reading Group: Distributed Reset (1990, FSTTCS) by Arora and Gouda
Paper: Distributed Reset
A reset subsystem is designed that can be embedded in an arbitrary distributed system in order to allow the system processes to reset the system when necessary. Our design is layered, and comprises three main components: a leader election, a spanning tree construction, and a diffusing computation. Each of these components is self-stabilizing in the following sense: if the coordination between the up-processes in the system is ever lost (due to failures or repairs of processes and channels), then each component eventually reaches a state where coordination is regained. This capability makes our reset subsystem very robust: it can tolerate fail-stop failures and repairs of processes and channels, even when a reset is in progress.
00:09:04 A. Jesse Jiryu Davis: I read Murat's posts! 00:09:09 Bharat Naik: Yes! 00:09:53 Zeeshan Lakhani: http://muratbuffalo.blogspot.com/2021/02/foundational-distributed-systems-papers.html 00:10:16 A. Jesse Jiryu Davis: "muh-ROT duh-MEER-bosh" I think 00:11:40 b5: Rooted spanning treez 00:20:51 A. Jesse Jiryu Davis: Paul Borrill - could you tell us where you work and link us to your paper? 00:21:49 Brent Shambaugh: What's an appropriate initial state? I might need to look in the paper to see if there is answer. 00:23:05 Paul Borrill: https://itsabouttime.club/ 00:31:20 Lawrence: Just listening in, very interesting paper! Has implications for all sorts of distributed systems, social systems, organizational resets, and handling hostile takeovers! 00:40:37 Brent Shambaugh: This triggered a thought: "quantum walks in Gremlin". Stay tuned for the video of the presentation. https://arxiv.org/abs/1511.06278 00:41:08 Paul Borrill: Contact me on email: email@example.com for more details about the DOE Distributed Resilient Systems Project 00:42:38 b5: By preselect you mean before the process joins the network? 00:44:38 Brent Shambaugh: https://www.youtube.com/watch?v=qRoAInXxgtc "GraphDayTexas 2016 Talks: Dr. Marko A. Rodriguez Quantum Processes in Graph Computing" 00:46:56 Lawrence: This reminds me of an idea I had about wireless communication using state similarity signaling 00:49:06 Paul Borrill: Next Saturday: https://www2.eecs.berkeley.edu/Faculty/Homepages/lee.html 00:49:26 Paul Borrill: Reactors enable determinism while preserving the style of actors, and without the performance loss of indiscriminately constraining concurrency to fix bugs due to race conditions and other sources of nondeterminism. Reactors promote modularity and allow for distributed execution. 00:51:42 Lawrence: could you hash the topological space and use that as the routing method? that way you can direct a packet based on the topological shape of the route? 00:53:40 b5: This will anger your router 😄 00:56:38 b5: @Lawrence, as soon as you hash something, you’re transitioning into an overlay network. I just finished a rundown on routing in an overlay network that you might find useful: https://www.youtube.com/watch?v=1QdKhNpsj8M 00:56:58 Lawrence: Reacted to "@Lawrence, as soon a..." with 👍 00:58:29 Paul Borrill: https://smartnicssummit.com 01:01:10 b5: @Paul, this one? https://wasl.uwaterloo.ca/projects/nifty/ 01:01:15 James Walker (@walkah): https://www.causalislands.com/ 01:01:47 Paul Borrill: Yes 01:01:50 b5: So pumped for causal islands
distributed_reset_transcript.pdf (79.6 KB)