CS Other Presentations

Department of Computer Science - University of Cyprus

Besides Colloquiums, the Department of Computer Science at the University of Cyprus also holds Other Presentations (Research Seminars, PhD Defenses, Short Term Courses, Demonstrations, etc.). These presentations are given by scientists who aim to present preliminary results of their research work and/or other technical material. Other Presentations serve as a forum for educating Computer Science students and related announcements are disseminated to the Department of Computer Science (i.e., the csall list):

RSS Directions

Presentations Coordinator: Demetris Zeinalipour

PhD Defense: Dependable Distributed Shared Memory Suitable for Large, Strongly Consistent Objects, Mrs. Andria Trigeorgi (University of Cyprus, Cyprus), Thursday, 3 April, 2025, 12:00-13:00 EET.

The Department of Computer Science at the University of Cyprus cordially invites you to the PhD Defense entitled:

Dependable Distributed Shared Memory Suitable for Large, Strongly Consistent Objects

Abstract:
EMC Digital Universe (https://www.cycloneinteractive.com/our-work/ emc-digital-universe/) estimates that the data in the digital universe doubles every two years. The International Data Corporation (IDC) (https://www.idc. com/) reports an increase from 26 Zettabytes (ZB) in 2017 to 64.2 ZB in 2020, and an estimated 175 ZB by 2025. To cope with this explosion of data, large infrastructures are needed. A Distributed Storage System (DSS) is an infrastructure that can split data across multiple servers. The autonomous components (users and storage nodes) in this distributed environment must decide how to modify the data in a way that ensures consistency while attempting to handle as many read/write requests concurrently as possible. Atomic consistency, the strongest consistency model (also known as linearizability), simplifies application development and facilitates concurrency reasoning, but it is challenging to implement effectively in a distributed setting without sacrificing performance. Due to this difficulty, many commercial DSSs avoid strong consistency. In research institutions, a plethora of algorithmic solutions along with proven correctness guarantees have been proposed to provide Atomic Distributed Shared Memory (ADSM) in a message passing system. ADSM provides the illusion of a sequential memory space despite asynchrony, network perturbations, and device failures. It is important to provide the necessary abstractions required for high performance at large scale in atomic algorithms, otherwise the concurrency reasoning provided by atomicity will be constrained by poor data scalability. This thesis aims at demonstrating that it is possible to build an ADSM with provable guarantees while providing highly concurrent access to users. A core of this work is the design and implementation of CoBFS, a distributed storage framework optimized for large-scale data. A key innovation in this framework is the introduction of fragmented objects, which enhance concurrency and reduce latency by enabling fine-grained read/write operations. at the block level. CoBFS adopts a modular architecture with two components: a Distributed Shared Memory Module (DSMM) to ensure consistency and a Fragmentation Module (FM) to manage the fragmentation strategy. To provide linearizable consistency at the DSMM, we incorporate CoABD, a coverable variant of the well-known ABD algorithm. This thesis formalizes and proves the consistency model of fragmented coverable linearizability tailored to support the operations enabled by fragmented objects. Additionally, we address fault tolerance and reconfigurability through the development of Ares, a dynamically reconfigurable storage algorithm that allows seamless node replacement and optimized performance under high concurrency. Comparative evaluations of Ares against commercial solutions such as Cassandra and Redis, reveal that Ares delivers similar or superior performance. Building on these contributions, the thesis integrates Ares into the fragmented strategy of CoBFS, resulting in CoAresF, a system capable of managing fragmented, erasure-coded objects with enhanced scalability and fault tolerance. To further optimize performance, we use distributed tracing tools to identify performance bottlenecks and develop of an optimized solution with techniques like batched operations, garbage collection, and piggybacking configuration data. The correctness of all algorithm designs in this thesis is formally proven. Furthermore, all implementations are evaluated through extensive experiments conducted on emulation and overlay testbeds, demonstrating their effectiveness across various scenarios.

Short Bio:
Andria Trigeorgi is a PhD candidate in Computer Science at the University of Cyprus (UCY), under the supervision of Prof. Chryssis Georgiou. She began her PhD in January 2020 and holds an MSc (2018) and a BSc (2016) in Computer Science from UCY. Her research focuses on Distributed Computing, with an emphasis on designing and implementing emulations of atomic read/write shared objects in message-passing systems, deployed on platforms such as Emulab, Grid5000, and AWS. Andria also works at Algolysis Ltd, led by Dr. Nicolas Nicolaou, collaborating with her supervisor on projects such as MA(R)S, CHARISMA, ARES-NGI, AM-NVE, and COLLABORATE. In addition, she has authored scientific publications and presented her work at international conferences, including SRDS’24, SSS’22, DISC’22, and SIROCCO’21, as well as in the journal ACM TOS. More at https://www.cs.ucy.ac.cy/~atrige01/

	Other Presentations Web: https://www.cs.ucy.ac.cy/colloquium/presentations.php
	Colloquia Web: https://www.cs.ucy.ac.cy/colloquium/
	Calendar: https://www.cs.ucy.ac.cy/colloquium/schedule/cs.ucy.pres.2025.trigeorgi.ics