Cloud Computing

Sponsors

IBM Ceská republika s.r.o.

Cisco Research

Cesnet

Team

Lukáš Kencl (Project Leader), Pavol Mulinka, OndĹ™ej Tománek

We conduct research that addresses two primary cloud-computing challenges: cloud data security and cloud latency. With the trend of cloud computing booming, companies are considering moving their applications to the cloud infrastructure, hoping to reduce overall operational costs. The cloud services market is growing fast but not so do the service provider offerings, which often vary only in terms of guaranteed storage capacity and bandwidth. Aspects such as data security or performance guarantees often remain obscure and may result in non-negligible additional costs or vulnerabilities.

Cloud Data Protection

Remote data storage services are very popular nowadays, be it cloud or various file-sharing websites. The problem of ensuring security in these services is imminent and even though many proprietary solutions exist, there is still no generally accepted cost-effective solution providing sufficient protection to the data in these storages. Strong semantically secure client-side encryption provides "perfect" security but prevents usage of any space-saving techniques and is thus extremely expensive for the provider. In this work, we investigate schemes allowing some data processing or reduction, while also providing a considerable security for the data. This work is in collaboration with IBM Research - Zurich.

Cloud Latency

Data-handling latency remains often a mission-critical application requirement and also a vital user-satisfaction parameter, yet its origins and means of control in the cloud-computing scenario remain poorly analyzed and understood. This research will take a deep look in the entire cloud-computing distributed system and analyze its elements. We employ theoretical concepts from the fields of computer networking and apply them to build a model for yielding optimized cloud network architectures and protocols.

Our goal is to take a measurements-based approach by constructing a cloud monitoring platform. A centralized monitor, residing in RDC network, collects data and generates statistics. Distributed probes capture raw traffic information and send them to the monitor. Our vision is to create a measurements-based cloud-latency model which can subsequently be optimized.

Publications

Technical Reports

Slides

[ ABSTRACT ] | [ SLIDES ]

[ ABSTRACT ] | [ SLIDES ]