Dr Dimitrios Pezaros, CEng, SMIEEE
Tel: +44 (0) 141 3306051
Fax: +44 (0) 141 3304913
Dr Dimitrios Pezaros is Senior Lecturer in the School of Computing Science at the University of Glasgow. He is a member of the Glasgow Systems Section (GLASS) and founding director of the Networked Systems Research Laboratory (netlab) at Glasgow. Dr Pezaros has published widely and is leading research in computer communications, network and service management, resilience and accoutnability of future virtualised networked infrastructures, exploring technologies such as, e.g., Software-Defined Networking (SDN) and Network Funciton Virtualization (NFV).
He has received funding in excess of £2.5m for his research by the Engineering and Physical Sciences Research Council (EPSRC), the European Comission (EC), the London Mathematical Society (LMS), the US Federal Aviation Administration (FAA), the University of Glasgow, and the industry (incl. Airbus Group, Brocade, Google, NATS, Solarflare, etc.).
Prior to joining Glasgow, Dr Pezaros has held postdoctoral and senior research associate positions at Lancaster University, where he has worked on a number of EPSRC and EU-funded projects in the areas of network measurement and management, traffic engineering, autonomic communications, and network resilience. He holds a BSc (Hons.) and a PhD in Computer Science from Lancaster University, and has been a doctoral fellow of Agilent Technologies Inc. between 2000 and 2004. He is a chartered engineer, a fellow of the HEA, and a senior member of the IEEE.
Visit our recently launched
Networked Systems Research Laboratory
Our paper on
SDN-based virtual machine
management for cloud data centers has won the best paper award at
IEEE CloudNet 2014
The Glasgow Raspberry Pi Cloud makes the news: Lego helps computing science students build future careers
Our paper on A two-level Markov model for packet loss in UDP/IP-based real-time video applications targeting residential users will appear in Elsevier Computer Networks
Our paper on Using programmable data networks to detect critical infrastructure challenges has appeared in CRITIS 2014
Our paper on Scalable traffic-aware virtual machine management for cloud data centers has appeared in ICDCS 2014 (acceptance rate 13%)
Our paper on Implementing Scalable, Network-Aware Virtual Machine Migration for Cloud Data Centers will appear in IEEE CLOUD 2013.
Our paper on Longer is Better: Exploiting Path Diversity in Data Center Networks will appear in ICDCS 2013 (acceptance rate 13%).
Our paper on The Glasgow Raspberry Pi Cloud: A Scale Model for Cloud Computing Infrastructures will appear in CCRM 2013.
Our paper on Internet Traffic Classification using Energy Time-Frequency Distributions will appear in IEEE ICC 2013.
Our poster on User-level Data Center Tomography has appeared in ACM SIGCOMM 2012.
Our paper on A Flexible and Open-Source Broadband Mapping Framework has appeared in IFIP/TC6 Networking 2012.
I have taught courses on Communications, Programming, Networked and Distributed Systems, and Network Measurement. Currently, I am teaching the following courses at Glasgow:
PDM1: Programming Digital Media (Spring 2011, 2012)
I am interested in supervising undergraduate and postgraduate projects in networked and distributed systems. Students with interests and ideas that lie in these broad areas should feel free to contact me.
I am interested in and have worked on a number of areas that can be embraced under the broad themes of network instrumentation, management, and performance evaluation. For a number of years, I have focused on network performance measurement using traffic instrumentation in order to reveal the actual performance delivered by the infrastructure to the diverse set of application flows routed over the Internet. Lately, I am interested in how instrumentation and network control can be exploited to enable a number of emergent network properties such as resilience and self-optimisation. The following summarise my main areas of interest:
Cloud/DataCentre Computing Optimisation and Control
Cloud and Data Centre services are rapidly emerging to replace Desktop computing as we know it and move storage and processing resources in the network. Building cloud environments following the best-effort Internet principles is unsustainable since providers need to maximise return on investment through resource usage efficiency rather than massive over-provisioning. I am interested in investigating instrumentation, control and routing algorithms and systems to manage the operation of cloud infrastructures for future services that will require performance and resource guarantees.
Network Measurement, Control and Management
Investigate the (partial) integration of measurement and control functionality with the Internet's main forwarding mechanism. Traditionally, the Internet has adopted the data/control plane (fast/slow path) separation following the legacy telecommunication network paradigm. However, we can only build self-managed and self-optimised networks if we can automate network control based on the temporal performance experienced by the operational network traffic. I am interested in the always-on operation of measurement and monitoring mechanisms, and their integration with traffic engineering network control structures such as routing.
Resilience is an emergent property of networked systems that reflects their ability to self-protect their own operation by detecting and defending against challenges that can have either a malicious or a legitimate intent. Similar to availability, resilience is an always-on property that is most important when the network is under stress. I am interested in investigating always-on mechanisms that use spare hardware or software processing capacity in network nodes to facilitate self-protection.
Network Traffic Modelling and Anomaly Detection
Analysis and modelling of network traffic has received significant attention from the research community during the past fifteen years. However, the fractal nature of network traffic that manifests itself at all scales in space and time, makes modelling of network traffic particularly challenging. I am looking into more tractable models to describe network traffic and characterise it under a probabilistic framework. At the same time, network anomalies become increasingly popular and significant research effort is being devoted in their characterisation. I am looking into methods for characterising anomalies accurately and at various levels of granularities, and also into methods for linking higher-level (anomalous) events with low-level manifestations.
QoS/QoE of Multimedia Network Protocols
Streaming of networked multimedia content is becoming increasingly popular partly due to the broadband speeds that have become commodity on residential network lines. However, before high-definition streaming content can replace traditional TV/Cable services, it is essential to understand the performance experienced by multimedia traffic over common technologies (e.g. DSL/Cable) that offer residential connectivity, and build protocols and management systems that can dynamically improve a range of Quality of Service (QoS) and Quality of Experience (QoE) parameters.
Home Network Management
Home networks are becoming increasingly complex since they integrate a diverse range of devices (e.g. PCs, smartphones, Set-Top Boxes, TVs, etc.) and most importantly are managed by inexperienced users. We are currently investigating mechanisms that offer greater visibility into the operation and dynamics of the home networking environment, and also enable the effective management and prioritisation of resources among the users of the home network.
I am looking for strong students interested in pursuing a PhD in the above areas.
I have been privileged to work with the following people:
Abeer Ali (PhD student, University of Glasgow)
Gubran Al-Kubati (PhD student, University of Glasgow)
Richard Cziva (PhD student and research assistant, University of Glasgow)
Simon Jouet (PhD student and research assistant, University of Glasgow)
Georgios Sfakianakis (PhD student, University of Glasgow)
Kyle White (PhD student, University of Glasgow)
Former students and researchers:
Dr Martin Ellis (PhD; now with Skype)
Gregg Hamilton (MRes; now with Linn)
Dr Angelos Marnerides (PhD/PDRA; now Lecturer, Liverpool John Moores University)
Dr Posco Tso (SICSA Research fellow; now Lecturer, Liverpool John Moores University)
Recent and Selected Publications
Jouet, S., Cziva, R., and Pezaros, D. (2015) Arbitrary Packet Matching in OpenFlow. In: 16th International Conference on High Performance Switching and Routing (HPSR), Budapest, Hungary, 1-4 Jul 2015. [Recipient of the best paper award]
Cziva, R., Stapleton, D., Tso, F. P., and Pezaros, D. P. (2014) SDN-based virtual machine management for cloud data centers. In: Third IEEE International Conference on Cloud Networking (IEEE CloudNet), 8-10 Oct 2014, Luxembourg. [Recipient of the best paper award]
Ellis, M., Pezaros, D. P., Kypraios, T., and Perkins, C. (2014) A two-level Markov model for packet loss in UDP/IP-based real-time video applications targeting residential users. Computer Networks, 70 . pp. 384-399. ISSN 1389-1286 (doi:10.1016/j.comnet.2014.05.013)
Tso, F. P., Oikonomou, K., Kavvadia, E., and Pezaros, D. P. (2014) Scalable traffic-aware virtual machine management for cloud data centers. In: IEEE International Conference on Distributed Computing Systems (IEEE ICDCS), 30 Jun - 3 Jul 2014, Madrid, Spain.
Tso, F.P., and Pezaros, D.P. (2013) Improving data centre network utilisation using near-optimal traffic engineering. IEEE Transactions on Parallel and Distributed Systems, 24 (6). pp. 1139-1148. ISSN 1045-9219 (doi:10.1109/TPDS.2012.343)
Marnerides, A., Pezaros, D., Kim, H.-C., and Hutchison, D. (2013) Internet traffic classification using energy time-frequency distributions. In: IEEE International Conference on Communications (ICC), 9-13 Jun 2013, Budapest, Hungary.
Tso, F.P., Hamilton, G., Weber, R., Perkins, C.S., and Pezaros, D.P. (2013) Longer is better: exploiting path diversity in data center networks. In: IEEE International Conference on Distributed Computing Systems (ICDCS) , 8-11 Jul 2013, Philadelphia, PA, USA.
Pezaros, D.P., Hoerdt, M., and Hutchison, D. (2011) Low-overhead end-to-end performance measurement for next generation networks. IEEE Transactions on Network and Service Management, 8 (1). pp. 1-14. (doi:10.1109/TNSM.2011.032311.090369 )
Pezaros, D., Georgopoulos, K., and Hutchison, D. (2010) High-speed, in-band performance measurement instrumentation for next generation IP networks. Computer Networks, 54 (18). pp. 3246-3263. ISSN 1389-1286 (doi:10.1016/j.comnet.2010.06.014)
Pezaros, D., Sifalakis, M., and Hutchison, D. (2007) On the long-range dependent behaviour of unidirectional packet delay of wireless traffic. In: IEEE Global Telecommunications Conference (IEEE GLOBECOM), November 26-30, 2007, Washington, DC.
Pezaros, D., Sifalakis, M., Schmid, S., and Hutchison, D. (2004) Dynamic link measurements using active components. In: Sixth International Working Conference on Active Networking (IWAN'04), October 27-29, 2004, Kansas, USA. [Recipient of the best paper award]
Pezaros, D., P., Network Traffic Measurement for the Next Generation Internet, PhD Thesis, Lancaster University, August 2005. [pdf's]