NavTalks

From Navigators

(Difference between revisions)
Jump to: navigation, search
Line 33: Line 33:
<tr>
<tr>
             <td style="width:10%">4 October</td>
             <td style="width:10%">4 October</td>
-
             <td style="width:30%"> Marcus Völp (Research Scientist, CritiX, SnT, Univ. of Luxembourg) </td>
+
             <td style="width:30%">Marcus Völp (Research Scientist, CritiX, SnT, Univ. of Luxembourg) </td>
             <td style="width:50%"><span title="As you are well aware, many practical concerns in systems aiming at Byzantine fault and intrusion tolerance require reaching consensus in difficult situations. For example, to remain exhaustion safe, replacing permanently damaged replicas requires relocating the replicated functionality to a fresh set of spares, necessitating conensus on the new group of active replicas. While group membership protocols exists for this task, we are also aware of their limitations (faults in the adaptation infrastructure (recurring the problem in the servers implementing it), operation modes that cannot reach consensus (aka Cheap / ReBFT minimal mode), etc.) that make it extremely difficult (if not impossible) to perform these reconfigurations in a reliable manner. In this talk, I would like to give you an overview over some of the current (unsolved) research problems we work on in CritiX and which I would like to discuss with you while here. I would like to share my view on our hinge that in some of the above settings, there is still hidden an impossibility result, possibly rendering CheapBFT (or at least generalizations of it to arbitrary quorums) incorrect, but motivating a novel design principle, which we call reflective consensus: Rather than solving the difficult, but naturally arising consensus problem (e.g., consensus on group membership in case of exhaustion failure due to an increasing threat level), we reflect consensus to the same set of replicas where it will occur, but in a simpler version that is possibly even executed at a different time (e.g., proactively when the system is not yet exhaustion failed). ">Reflective Consensus</span></td>
             <td style="width:50%"><span title="As you are well aware, many practical concerns in systems aiming at Byzantine fault and intrusion tolerance require reaching consensus in difficult situations. For example, to remain exhaustion safe, replacing permanently damaged replicas requires relocating the replicated functionality to a fresh set of spares, necessitating conensus on the new group of active replicas. While group membership protocols exists for this task, we are also aware of their limitations (faults in the adaptation infrastructure (recurring the problem in the servers implementing it), operation modes that cannot reach consensus (aka Cheap / ReBFT minimal mode), etc.) that make it extremely difficult (if not impossible) to perform these reconfigurations in a reliable manner. In this talk, I would like to give you an overview over some of the current (unsolved) research problems we work on in CritiX and which I would like to discuss with you while here. I would like to share my view on our hinge that in some of the above settings, there is still hidden an impossibility result, possibly rendering CheapBFT (or at least generalizations of it to arbitrary quorums) incorrect, but motivating a novel design principle, which we call reflective consensus: Rather than solving the difficult, but naturally arising consensus problem (e.g., consensus on group membership in case of exhaustion failure due to an increasing threat level), we reflect consensus to the same set of replicas where it will occur, but in a simpler version that is possibly even executed at a different time (e.g., proactively when the system is not yet exhaustion failed). ">Reflective Consensus</span></td>
             <td style="width:10%">&nbsp;</td>
             <td style="width:10%">&nbsp;</td>
Line 39: Line 39:
<tr>
<tr>
             <td style="width:10%">18 October</td>
             <td style="width:10%">18 October</td>
-
             <td style="width:30%"> Yair Amir (Professor, Johns Hopkins University) </td>
+
             <td style="width:30%">Yair Amir (Professor, Johns Hopkins University) </td>
-
             <td style="width:50%"> Timely, Reliable, and Cost-Effective Internet Transport Service using Structured Overlay Networks</td>
+
             <td style="width:50%"><span title="Emerging applications such as remote manipulation, collaborative virtual reality, or remote robotic surgery require communication that is both timely and reliable, but the Internet natively supports only communication that is either completely reliable with no timeliness guarantees (e.g. TCP) or timely with only best-effort reliability (e.g. UDP). We present an overlay transport service that can provide highly reliable communication while meeting stringent timeliness guarantees (e.g. 130ms round-trip latency across the US) over the Internet.
 +
 
 +
To enable routing schemes that can support the necessary timeliness and reliability, we introduce dissemination graphs, providing a unified framework for specifying routing schemes ranging from a single path, to multiple disjoint paths, to arbitrary graphs. Based on an extensive analysis of real-world network data, we develop a timely dissemination-graph-based routing method that can add targeted redundancy in problematic areas of the network. We show that this approach can cover close to 99% of the performance gap between a traditional single-path approach and an optimal (but prohibitively expensive) scheme.">Timely, Reliable, and Cost-Effective Internet Transport Service using Structured Overlay Networks</span></td>
             <td style="width:10%">&nbsp;</td>
             <td style="width:10%">&nbsp;</td>
         </tr>
         </tr>

Revision as of 11:38, 12 November 2018

The Navtalks is a series of informal talks given by Navigators members or some special guests about every two-weeks at Ciências, ULisboa.

Leave mouse over title's presentation to read the abstract.

Contents

September 2018

20 September Alysson Bessani SMaRtChain: A Principled Design for a New Generation of Blockchains  
20 September Rui Miguel Named Data Networking with Programmable Switches  

October 2018

4 October Bruno Vavala (Research Scientist in Intel Labs) Private Data Objects  
4 October Marcus Völp (Research Scientist, CritiX, SnT, Univ. of Luxembourg) Reflective Consensus  
18 October Yair Amir (Professor, Johns Hopkins University) Timely, Reliable, and Cost-Effective Internet Transport Service using Structured Overlay Networks  


November 2018

13/11 Salvatore Signorello The Past, the Present and some Future of Interest Flooding Attacks in Named-Data Networking  
13/11 Tiago Oliveira Vawlt - Privacy-Centered Cloud Storage  
27/11 Nuno Neves    
27/11 Ricardo Mendes    

December 2018

11/12 António Casimiro    
11/12Carlos Nascimento      

January 2019

15/01 Fernando Alves    
15/01 Ibéria Medeiros    
29/01 Fernando Ramos    
29/01 Miguel Garcia    

February 2019

19/02 Ana Fidalgo    
19/02 João Sousa    

March 2019

12/03 Pedro Gaspar    
12/03 Ricardo Morgado    
26/03 André Oliveira    
26/03 Nuno Dionísio    

April 2019

09/04 Adriano Serckumecka    
09/04 Tulio Ribeiro    
30/04 Miguel Moreira    
30/04 Pedro Ferreira    

May 2019

14/05 Diogo Gonçalves    
14/05 Vinicius Cogo    
28/05 Francisco Araújo    
28/05 Miguel Matos    

June 2019

11/06 Eric Vial    
11/06 Robin Vassantlal    
25/06 João Pinto    
25/06 Tiago Correia    
Personal tools
Navigators toolbox