Tid: 2022-09-30 14:15 till 15:00
Plats: E:2116
Kontakt: birger [dot] swahn [at] cs [dot] lth [dot] se
Spara händelsen till din kalender

One Computer Science MSc thesis to be presented on 30 September

Friday, 30 September there will be a master thesis presentation in Computer Science at Lund University, Faculty of Engineering.

The presentation will take place in E:2116.

Note to potential opponents: Register as an opponent to the presentation of your choice by sending an email to the examiner for that presentation (firstname.lastname@cs.lth.se). Do not forget to specify the presentation you register for! Note that the number of opponents may be limited (often to two), so you might be forced to choose another presentation if you register too late. Registrations are individual, just as the oppositions are! More instructions are found on this page.

14:15-15:00 in E:2116

Presenter: Anna Sjerling
Title: Preserving availability in a consensus module using back pressure
Examiner: Flavius Gruian
Supervisors: Emma Söderberg (LTH), Ragnar Wernersson (Neo4j)

In distributed systems, the consensus algorithm Raft is used to replicate a globally ordered log of entries. Members that fall behind in replicating the log entries can cause system write unavailability. One reason for this write unavailability is that Raft needs a majority of members to replicate a log entry, before it is accepted into the system. The purpose of this Master's Thesis Project is to investigate and design a back pressure solution which would prevent members from falling behind, and thus reduce this form of unavailability. To understand the problem better, we preformed experiments to investigate which parameters effected the occurrence of members falling behind. The experiments were preformed at Neo4j, the company where this Master's Thesis Project was carried out. To investigate possible solutions, we conducted a literature review where solutions to similar problems were considered. Based on the existing solutions, we made a theoretical evaluation of possible solutions. The experiments that investigated parameters CPU+RAM, IOPS and transaction size, indicated that all three parameters had an impact on the occurrence of members falling behind. We present the possible solutions as a three dimensional design space, which theoretically construct 27 different solutions, one of which was implemented and provided proof of concept. This made us conclude that back pressure can be implemented in a system which uses Raft, to better avoid write unavailability caused by members falling behind in replicating the globally ordered log.

Link to popular science summary: To be updated