ΕΛ EN
Home | People | Georgiou, Michael
5
Scholarly Works
40%
OA Ratio
40%
Collaborative Ratio
Scholar

Georgiou, Michael

Department of Electrical Engineering, Computer Engineering and Informatics

3
H-index (Scopus)
52
Scholar Citations

Scholarly Works

Research Output

Recent Publications

Hihooi: A Database Replication Middleware for Scaling Transactional Databases Consistently

Sirivianos M., Herodotou H., Paphitis A., Georgiou M.A.
Publishing year: 2022
Journal article
IEEE Transactions on Knowledge and Data Engineering, 2022, Volume 34, Issue 2
Scopus
Pages 691-707
With the advent of the Internet and Internet-connected devices, modern business applications can experience rapid increases as well as variability in transactional workloads. Database replication has been employed to scale performance and improve availability of relational databases but past approaches have suffered from various issues including limited scalability, performance versus consistency tradeoffs, and requirements for database or application modifications. This paper presents Hihooi, a replication-based middleware system that is able to achieve workload scalability, strong consistency guarantees, and elasticity for existing transactional databases at a low cost. A novel replication algorithm enables Hihooi to propagate database modifications asynchronously to all replicas at high speeds, while ensuring that all replicas are consistent. At the same time, a fine-grained routing algorithm is used to load balance incoming transactions to available replicas in a consistent way. Our thorough experimental evaluation with several well-established benchmarks shows how Hihooi is able to achieve almost linear workload scalability for transactional databases.

Attaining Workload Scalability and Strong Consistency for Replicated Databases with Hihooi

Sirivianos M., Herodotou H., Paphitis A., Georgiou M.A., Odysseos L., Panayiotou M.
Publishing year: 2021
Conference paper
Proceedings of the ACM SIGMOD International Conference on Management of Data, 2021
Scopus
Pages 2721-2725
Database replication can be employed for scaling transactional workloads while maintaining strong consistency semantics. However, past approaches suffer from various issues such as limited scalability, performance versus consistency tradeoffs, and requirements for database or application modifications. Hihooi is a new replication-based master-slave middleware system that is able to overcome the aforementioned limitations. The novelty of Hihooi lies in its modern architecture as well as its replication and transaction routing algorithms. In particular, Hihooi replicates all write statements asynchronously and applies them in parallel at the replica nodes, while ensuring replica consistency. At the same time, a fine-grained transaction routing algorithm ensures that all read transactions are load balanced to the replicas consistently. This demonstration will showcase the key functionalities of Hihooi, including (i) practical management of system components and databases (e.g., add a new replica node), (ii) increased scalability compared to state-of-the-art approaches, and (iii) support for elasticity by suspending and resuming database replicas online without service interruption.

Towards auto-scaling existing transactional databases with strong consistency

Sirivianos M., Herodotou H., Paphitis A., Georgiou M.A.
Publishing year: 2019
Conference paper
Proceedings 2019 IEEE 35th International Conference on Data Engineering Workshops Icdew 2019, 2019
Scopus
Pages 107-112
Existing relational database systems often suffer from rapid increases or significant variability of transactional workloads but lack support for scalability or elasticity. Database replication has been employed to scale workload performance but past approaches make various performance versus consistency tradeoffs and typically lack the mechanisms and policies for dynamically adding and removing replicas. This paper presents Hihooi, a replication-based middleware system that is able to achieve scalability, strong consistency, and elasticity for existing transactional databases. These features are enabled by (i) a novel replication algorithm for propagating database modifications asynchronously and consistently to all replicas at high speeds, and (ii) a new routing algorithm for directing incoming transactions to consistent replicas. Our experimental evaluation validates the high scalability and elasticity benefits offered by Hihooi, which form the key ingredients towards a truly auto-scaling system.

SLA information management through dependency digraphs: The case of cloud data services

Georgiou, Michael, Kantere, Verena M., Morin, Jean Henry, Stamou, Katerina
Publishing year: 2014
Conference paper Repository
The paper addresses the issue of Service Level Agreement (SLA) data management for the cloud computing domain. We discuss the SLA anatomy and provide an analysis of SLA data management requirements. The analysis highlights SLA data aspects considering the association of SLA terms with service management operations. Our work proposes a simple, structured way to store and manage SLA information through a digraph data model that is modular, extensible and expressive with respect to data operational dependencies. The proposed SLA digraph considers properties of SLA terms and service component dependencies. We sketch a realistic dataservice provisioning scenario, where the proposed SLA digraph is applied. We illustrate the mapping of data service attributes into SLA terms and the role of edge properties in the definition of service dependencies.

A SLA graph model for data services

Georgiou, Michael, Kantere, Verena M., Morin, Jean Henry, Stamou, Katerina
Publishing year: 2013
Conference paper Repository
Cloud computing has given rise to on-demand service provisioning and massive outsourcing of IT infrastructures and applications to virtual, commoditized ones. Despite the broad Service Level Agreement (SLA) usage in scientific settings, their role in cloud markets is peripheral and misinterpreted. The paper introduces a SLA graph data model that supports automated SLA formalization and data management through a property digraph. The data model is described as a directed graph (digraph). We elaborate on node and edge properties that indicate dependencies in the SLA data management flow. We sketch a realistic scenario of cloud data service provisioning to extract attributes that characterize the data service. The SLA graph model and data service attributes are used to demonstrate the formalization of a SLA template that is managed as a property graph. The graph structure enables the manipulation of SLA information in a modular, extensible way that considers the data flow and all inclusive data dependencies.

5
Unique publications
40%
Open access (2 publications; 0 without information)
40%
With other institutions (2 publications; 0 without information)

H-Index Reports

Index type Scopus Google Scholar Web of Science
H-index 3 4 0
H5-index 1 1 0
H10-index 2 4 0

Publications per year and source

No description yet.
Faculty
Faculty of Engineering and Technology
Department
Department of Electrical Engineering, Computer Engineering and Informatics