香港六合彩

Dr Georgios听Zervas, senior lecturer in Electronic and Electrical Engineering, along with members of ICCS, this week head to Cyprus to present work on future-proofed networks for Cloud Data Centres.

The researchers will deliver presentations to the international Photonics in Switching and Computing Conference听on two research papers along with an invited talk to a workshop co-located with the conference.

George Zervas will听give the invited talk to those researching and developing infrastructures听for Cloud Data centres of the future, specifically looking at how the Data Centre landscape moves beyond the limitation being imposed by the slow down of Moore's Law.

Dr Zervas said:

鈥�We are challenging many of the long-standing architectures and principles for data-centre inrfastructure. The current methods deployed in data centres will not keep up with scalability and performance required of future Cloud Data Centres.

The talk will elaborate on the听shortcoming of existing听technologies in computing and networking, identifying the challenges yet to come. Further, it will highlight the solutions being developed at 香港六合彩 which bring together network science and hardware technologies towards scalable optical networks that are deterministic and able to operate in听nanoseconds.

In addition to the invited talk, the two research papers being presented within the main conference will showcase the groundbreaking work of the group.听

Paris Andreades, PhD student听within the 香港六合彩 Optical Networks Group and lead researcher for the first paper, will present a leap in the current state-of-the-art for data centre networking. The paper presents a new method of controlling the paths through which data will flow听in the network, known as switching. This novel design can enable switching of these paths at a nanosecond time-scale and support a significant increase in the number of data inputs and outputs within a network without a decrease in performance speed.

Joshua Benjamin, Research Assistant and PhD Student in the Optical Networks Group, led the work presented in the second paper. He will report on a听ten-fold increase in data throughput of the network made possible by a听novel network architecture and control method. Investigations have听shown this method could potentially be scaled over 16 times more servers, dramatically increasing the network's capability.

Further explaining the wider group's work Dr Zervas said:听

The work we are presenting at PSC is part of on-going research that has seen the development of a unique network that can support 10,000s to potentially 100,000s computing nodes but does not require switiching or routing within the network making a data-centre network that is reliable at lower latency, increasing the speed at which data can be processed.听

Explaining the magnitude of current innovations听he added:

This is a fascinating period of work that has been made possible by ground-breaking developments in network science, hardware acceleration methods and optical networking.鈥�