Smart execution of distributed application by balancing resources in mobile devices and cloud-based avatars

L’obiettivo del progetto di tesi svolto è quello di realizzare un servizio di livello middleware dedicato ai dispositivi mobili che sia in grado di fornire il supporto per l’offloading di codice verso una infrastruttura cloud. In particolare il progetto si concentra sulla migrazione di codice verso macchine virtuali dedicate al singolo utente. Il sistema operativo delle VMs è lo stesso utilizzato dal device mobile. Come i precedenti lavori sul computation offloading, il progetto di tesi deve garantire migliori performance in termini di tempo di esecuzione e utilizzo della batteria del dispositivo. In particolare l’obiettivo più ampio è quello di adattare il principio di computation offloading a un contesto di sistemi distribuiti mobili, migliorando non solo le performance del singolo device, ma l’esecuzione stessa dell’applicazione distribuita. Questo viene fatto tramite una gestione dinamica delle decisioni di offloading basata, non solo, sullo stato del device, ma anche sulla volontà e/o sullo stato degli altri utenti appartenenti allo stesso gruppo. Per esempio, un primo utente potrebbe influenzare le decisioni degli altri membri del gruppo specificando una determinata richiesta, come alta qualità delle informazioni, risposta rapida o basata su altre informazioni di alto livello. Il sistema fornisce ai programmatori un semplice strumento di definizione per poter creare nuove policy personalizzate e, quindi, specificare nuove regole di offloading. Per rendere il progetto accessibile ad un più ampio numero di sviluppatori gli strumenti forniti sono semplici e non richiedono specifiche conoscenze sulla tecnologia. Il sistema è stato poi testato per verificare le sue performance in termini di mecchanismi di offloading semplici. Successivamente, esso è stato anche sottoposto a dei test per verificare che la selezione di differenti policy, definite dal programmatore, portasse realmente a una ottimizzazione del parametro designato.

An Architecture to offer Cloud-Based Radio Access Network as a Service

This paper addresses the novel notion of offering a radio access network as a service. Its components may be instantiated on general purpose platforms with pooled resources (both radio and hardware ones) dimensioned on-demand, elastically and following the pay-per-use principle. A novel architecture is proposed that supports this concept. The architecture's success is in its modularity, well-defined functional elements and clean separation between operational and control functions. By moving much processing traditionally located in hardware for computation in the cloud, it allows the optimisation of hardware utilization and reduction of deployment and operation costs. It enables operators to upgrade their network as well as quickly deploy and adapt resources to demand. Also, new players may easily enter the market, permitting a virtual network operator to provide connectivity to its users.

A library-based bioinformatics services program*

Support for molecular biology researchers has been limited to traditional library resources and services in most academic health sciences libraries. The University of Washington Health Sciences Libraries have been providing specialized services to this user community since 1995. The library recruited a Ph.D. biologist to assess the molecular biological information needs of researchers and design strategies to enhance library resources and services. A survey of laboratory research groups identified areas of greatest need and led to the development of a three-pronged program: consultation, education, and resource development. Outcomes of this program include bioinformatics consultation services, library-based and graduate level courses, networking of sequence analysis tools, and a biological research Web site. Bioinformatics clients are drawn from diverse departments and include clinical researchers in need of tools that are not readily available outside of basic sciences laboratories. Evaluation and usage statistics indicate that researchers, regardless of departmental affiliation or position, require support to access molecular biology and genetics resources. Centralizing such services in the library is a natural synergy of interests and enhances the provision of traditional library resources. Successful implementation of a library-based bioinformatics program requires both subject-specific and library and information technology expertise.

Design and Implementation of a Cloud-based Middleware for Persuasive Android Applications

Negli ultimi decenni, le tecnologie e i prodotti informatici sono diventati pervasivi e sono ora una parte essenziale delle nostre vite. Ogni giorno ci influenzano in maniera più o meno esplicita, cambiando il nostro modo di vivere e i nostri comportamenti più o meno intenzionalmente. Tuttavia, i computer non nacquero inizialmente per persuadere: essi furono costruiti per gestire, calcolare, immagazzinare e recuperare dati. Non appena i computer si sono spostati dai laboratori di ricerca alla vita di tutti i giorni, sono però diventati sempre più persuasivi. Questa area di ricerca è chiamata pesuasive technology o captology, anche definita come lo studio dei sistemi informatici interattivi progettati per cambiare le attitudini e le abitudini delle persone. Nonostante il successo crescente delle tecnologie persuasive, sembra esserci una mancanza di framework sia teorici che pratici, che possano aiutare gli sviluppatori di applicazioni mobili a costruire applicazioni in grado di persuadere effettivamente gli utenti finali. Tuttavia, il lavoro condotto dal Professor Helal e dal Professor Lee al Persuasive Laboratory all’interno dell’University of Florida tenta di colmare questa lacuna. Infatti, hanno proposto un modello di persuasione semplice ma efficace, il quale può essere usato in maniera intuitiva da ingegneri o specialisti informatici. Inoltre, il Professor Helal e il Professor Lee hanno anche sviluppato Cicero, un middleware per dispositivi Android basato sul loro precedente modello, il quale può essere usato in modo molto semplice e veloce dagli sviluppatori per creare applicazioni persuasive. Il mio lavoro al centro di questa tesi progettuale si concentra sull’analisi del middleware appena descritto e, successivamente, sui miglioramenti e ampliamenti portati ad esso. I più importanti sono una nuova architettura di sensing, una nuova struttura basata sul cloud e un nuovo protocollo che permette di creare applicazioni specifiche per smartwatch.

Optimizing the data intensive mediator-based Web services composition

Effectively using heterogeneous, distributed information has attracted much research in recent years. Current web services technologies have been used successfully in some non data intensive distributed prototype systems. However, most of them can not work well in data intensive environment. This paper provides an infrastructure layer in data intensive environment for the effectively providing spatial information services by using the web services over the Internet. We extensively investigate and analyze the overhead of web services in data intensive environment, and propose some new optimization techniques which can greatly increase the system’s efficiency. Our experiments show that these techniques are suitable to data intensive environment. Finally, we present the requirement of these techniques for the information of web services over the Internet.

Computazione Embodied e Disembodied: Cloud-based IoT

La tesi esplora la co-esistenza di computazioni embodied e disembodied nei moderni sistemi software, adottando come caso di studio il recente trend che vede sempre più coesi e integrati sistemi per l'Internet of Things e sistemi Cloud-based. Si analizzano i principali modelli di comunicazione, protocolli di comunicazione e architetture situate. Inoltre si realizza una piattaforma IoT Middleware cloud-based per mostrare come la computazione possa essere distribuita lato embodied e disembodied.

BUILDING EFFICIENT AND COST-EFFECTIVE CLOUD-BASED BIG DATA MANAGEMENT SYSTEMS

In today’s big data world, data is being produced in massive volumes, at great velocity and from a variety of different sources such as mobile devices, sensors, a plethora of small devices hooked to the internet (Internet of Things), social networks, communication networks and many others. Interactive querying and large-scale analytics are being increasingly used to derive value out of this big data. A large portion of this data is being stored and processed in the Cloud due the several advantages provided by the Cloud such as scalability, elasticity, availability, low cost of ownership and the overall economies of scale. There is thus, a growing need for large-scale cloud-based data management systems that can support real-time ingest, storage and processing of large volumes of heterogeneous data. However, in the pay-as-you-go Cloud environment, the cost of analytics can grow linearly with the time and resources required. Reducing the cost of data analytics in the Cloud thus remains a primary challenge. In my dissertation research, I have focused on building efficient and cost-effective cloud-based data management systems for different application domains that are predominant in cloud computing environments. In the first part of my dissertation, I address the problem of reducing the cost of transactional workloads on relational databases to support database-as-a-service in the Cloud. The primary challenges in supporting such workloads include choosing how to partition the data across a large number of machines, minimizing the number of distributed transactions, providing high data availability, and tolerating failures gracefully. I have designed, built and evaluated SWORD, an end-to-end scalable online transaction processing system, that utilizes workload-aware data placement and replication to minimize the number of distributed transactions that incorporates a suite of novel techniques to significantly reduce the overheads incurred both during the initial placement of data, and during query execution at runtime. In the second part of my dissertation, I focus on sampling-based progressive analytics as a means to reduce the cost of data analytics in the relational domain. Sampling has been traditionally used by data scientists to get progressive answers to complex analytical tasks over large volumes of data. Typically, this involves manually extracting samples of increasing data size (progressive samples) for exploratory querying. This provides the data scientists with user control, repeatable semantics, and result provenance. However, such solutions result in tedious workflows that preclude the reuse of work across samples. On the other hand, existing approximate query processing systems report early results, but do not offer the above benefits for complex ad-hoc queries. I propose a new progressive data-parallel computation framework, NOW!, that provides support for progressive analytics over big data. In particular, NOW! enables progressive relational (SQL) query support in the Cloud using unique progress semantics that allow efficient and deterministic query processing over samples providing meaningful early results and provenance to data scientists. NOW! enables the provision of early results using significantly fewer resources thereby enabling a substantial reduction in the cost incurred during such analytics. Finally, I propose NSCALE, a system for efficient and cost-effective complex analytics on large-scale graph-structured data in the Cloud. The system is based on the key observation that a wide range of complex analysis tasks over graph data require processing and reasoning about a large number of multi-hop neighborhoods or subgraphs in the graph; examples include ego network analysis, motif counting in biological networks, finding social circles in social networks, personalized recommendations, link prediction, etc. These tasks are not well served by existing vertex-centric graph processing frameworks whose computation and execution models limit the user program to directly access the state of a single vertex, resulting in high execution overheads. Further, the lack of support for extracting the relevant portions of the graph that are of interest to an analysis task and loading it onto distributed memory leads to poor scalability. NSCALE allows users to write programs at the level of neighborhoods or subgraphs rather than at the level of vertices, and to declaratively specify the subgraphs of interest. It enables the efficient distributed execution of these neighborhood-centric complex analysis tasks over largescale graphs, while minimizing resource consumption and communication cost, thereby substantially reducing the overall cost of graph data analytics in the Cloud. The results of our extensive experimental evaluation of these prototypes with several real-world data sets and applications validate the effectiveness of our techniques which provide orders-of-magnitude reductions in the overheads of distributed data querying and analysis in the Cloud.

Blended learning and the flipped classroom: The affordances of cloud based, located, and virtual world environments to support student learning

In this paper the author reports on the introduction of the flipped classroom integrating located, online and virtual world learning environments to support the collaborative lived experiences of a group of students and the educator participating in a higher education undergraduate art unit, Navigating the Visual World. A qualitative narrative methodology, A/r/tography, incorporating both image making and textual recording is used to explore and identify interwoven aspects of the artist/ researcher/ educator relationship in the creative artistic process of exploring concepts of identity within inquiry based art practice. Selected student examples, including a collaborative group assessment project demonstrate effective student engagement with experiential blended learning within the flipped classroom.

Cloud-based teaching in an engineering-physics course

This paper presents a transition from passive, traditional delivery of teaching to an active, “cloud-based” method, in a freshman engineering-physics course. The course is delivered to a traditional on-campus cohort, and also to an offcampus cohort by means of distance education and online learning. Cloud teaching refers to delivering education by means of websites and mobile-technology applications, where constant student attendance at the host campus is not always necessary. This is contrasted with traditional on-campus teaching, which occurs in a classroom. The use of lectures has been reduced while the use of tutorial and lab classes has increased. The new course structure was delivered for the first time in 2014, has run for two semesters, and will continue in 2015. It was found that student performance in the new structure was no worse than that in the older structure. Off-campus students in general welcomed the changes, while on-campus satisfaction did not change from before to after the transition.

Providing fairer resource allocation for multi-tenant cloud-based systems

A fundamental premise in cloud computing is trying to provide a more sophisticated computing resource sharing capability. In order to provide better allocation, the Dominant Resource Fairness (DRF) approach has been developed to address the "fair resource allocation problem" at the application layer for multi-tenant cloud applications. Nevertheless conventional DRF only considers the interplay of CPU and memory, which may result in over allocation of resources to one tenant's application to the detriment of others. In this paper, we propose an improved DRF algorithm with 3-dimensional demand vector to support disk resources as the third dominant shared resource, enhancing fairer resource sharing. Our technique is integrated with LINUX 'group' controls resource utilisation and realises data isolation to avoid undesirable interactions between co-located tasks. Our method ensures all tenants receive system resources fairly, which improves overall utilisation and throughput as well as reducing traffic in an over-crowded system. We evaluate the performance of different types of workload using different algorithms and compare ours to the default algorithm. Results show an increase of 15% resource utilisation and a reduction of 59% completion time on average, indicating that our DRF algorithm provides a better, smoother, fairer high-performance resource allocation scheme for both continuous workloads and batch jobs.

From science gateways to commercial solutions

The European CloudSME project that incorporated 24 European SMEs, besides five academic partners, has finished its funded phase in March 2016. This presentation will provide a summary of the results of the project, and will analyze the challenges and differences when developing “SME Gateways”, when compared to “Science Gateways”. CloudSME started in 2013 with the aim to develop a cloud-based simulation platform for manufacturing and engineering SMEs. The project was based around industry use-cases, five of which were incorporated in the project from the start, and seven additional ones that were added as an outcome of an open call in January 2015. CloudSME utilized science gateway related technologies, such as the commercial CloudBroker Platform and the WS-PGRADE/gUSE Gateway Framework that were developed in the preceding SCI-BUS project. As most important outcome, the project successfully implemented 12 industry quality demonstrators that showcase how SMEs in the manufacturing and engineering sector can utilize cloud-based simulation services. Some of these solutions are already market-ready and currently being rolled out by the software vendor companies. Some others require further fine-tuning and the implementation of commercial interfaces before being put into the market. The CloudSME use-cases came from a very wide application spectrum. The project implemented, for example, an open marketplace for micro-breweries to optimize their production and distribution processes, an insole design validation service to be used by podiatrists and shoe manufacturers, a generic stock management solution for manufacturing SMEs, and also several “classical” high-performance computing case-studies, such as fluid dynamics simulations for model helicopter design, and dual-fuel internal combustion engine simulation. As the project generated significant impact and interest in the manufacturing sector, 10 CloudSME stakeholders established a follow-up company called CloudSME UG for the future commercialization of the results. Besides the success stories, this talk would also like to highlight the difficulties when transferring the outcomes of an academic research project to real commercial applications. The different mindset and approach of academic and industry partners presented a real challenge for the CloudSME project, with some interesting and valuable lessons learnt. The academic way of supporting SMEs did not always work well with the rather different working practices and culture of many participants. Also, the quality of support regarding operational solutions required by the SMEs is well beyond the typical support services academic institutions are prepared for. Finally, a clear lack of trust in academic solutions when compared to commercial solutions was also imminent. The talk will highlight some of these challenges underpinned by the implementation of the CloudSME use-cases.

Efficient Security and Privacy Enhancing Solutions in Untrusted Environments

Thesis (Ph.D.)--University of Washington, 2016-08