Most of the advanced microprocessors today incorporate on-chip hardware performance counters. These counters are capable to count various events in a non-invasive way, while executing real workloads. Events such as the number of instructions, memory accesses, cache and TLB misses are the most common ones that can be precisely measured. The primary accomplishment of this work was to implement a performance monitoring tool, which could be used to evaluate system behaviour on high-end processing platforms. The tool is able to collect data from hardware performance counters and present them in an interpretable way. Moreover, it has support for two different platforms and two operating systems. As a secondary objective, several measurements were carried out on both supported platforms and operating systems to demonstrate the tool’s capabilities, and to solve the potential use-cases. / De flesta av dagens mikroprocessorer innehåller prestandaräknare direkt i hårdvaran. Dessa räknare kan räkna olika typer av händelser på ett icke störande sätt medans hårdvaran är under last. Händelser såsom instruktioner, minnesaccesser, cache och TLB missar är de vanligast förekommande räknarna som kan göras precisa. Det främsta genomförandet i denna uppgift var att implementera att verktyg för att övervaka prestanda som kan användas för att beräkna ett systems beteende på högprestanda-plattformar. Verktyget kan hämta prestandaräknarna och presentera dem i ett läsbart format. Dessutom har verktyget stöd för två olika plattformar och det översattes till två olika operativsystem. Som ett sekundärt mål gjordes många mätningar på de båda plattformarna samt operativsystemen som stöddes för att visa verktygets funktion och lösa potentiella användningsfall.

Since 1960, when the rst slide projectors were manufactured and people started using slides as a support for delivering information to their audience in educational and institutional environments, not much has changed. Nowadays slide presentations are still based on the traditional interaction methods between presenter and audience that limit the experience to the live event. Considering the worldwide raising number of smartphone users and the increasing interest by researchers in mobile education, this research aims to give its contribution to a marginally explored field: the mobile presentation systems space. Mobile presentation system is intended to be a fully mobile service that supports presenter and audience during all the phases of a slide based event and enhance the user experience by enabling interaction between physical and digital presentation spaces. The first goal of this work was to understand if there is an underserved demand for innovation in the current presentation interaction model and what the perceived needs and wants are. People sensible to this field went through interviews and surveys, conducted from both the presenter and the audience perspectives. The results showed that about 87% of the participants had previously experienced visibility problems during presentations, causing frustration and lack of concentration damaging both audience and presenter. 80% of the people use email for sharing their slide set to the audience, a non effective way to reach all the attendants. The ability to send/get the material was considered critical both from the presenter and the audience points of view, assessing it 3.3/5 and 3.9/5 respectively. Other major problems of technical nature (software and hardware) emerged as often experienced by almost half the participants. This preliminary qualitative investigation built the basis of the PresentAction concept that was conceived to solve the raised problems and to support presenter and audience during all the phases of a presentation (before, during and after the live session) in a single solution. The concept was the guideline of the designed and developed system. PresentAction lets any connected smartphone user, regardless his/her technical skills, run slideshows from already existing slide decks anywhere and at any time and automatically create complete, high quality, interactive, updatable, learning object delivered to the attending users through smart content delivery methods. The system was experimentally validated, both technically (several performance measurements) and from the end user experience point of view (system trial and surveys). The users were really satisfied, rating the developed system at 4.25/5 and with more than 80% of the participants willing to adopt this system both as presenter and audience. The users emphasized in particular the usefulness of being able to run presentations even without the support of a projector (connecting directly presenter and audience devices) and the possibility of creating a personal, mobile portfolio of "talking" slides of the events they attended. The performed experiments also demonstrated that pre-fetching based slide distribution is fundamental for increasing the number of concurrent users in the audience while keeping a real time feeling for the presentation, with limited delays in the transition between slides on their mobile device. Finally a new concept of slides as learning multimedia objects, merging images with audio and supporting a variety of digital on demand services, was developed and demonstrated as a superior solution, in terms of both communication costs and user experience, to accessing a complete raw video of the presentation event. Experimental evidence showed that this approach can require as low as 945 times less communication resource than current video-based solutions.

The pay-as-you-go pricing model and the illusion of unlimited resources makes cloud computing a conducive environment for provision of elastic services where different resources are dynamically requested and released in response to changes in their demand. The benefit of elastic resource allocation to cloud systems is to minimize resource provisioning costs while meeting service level objectives (SLOs). With the emergence of elastic services, and more particularly elastic key-value stores, that can scale horizontally by adding/removing servers, organizations perceive potential in being able to reduce cost and complexity of large scale Web 2.0 applications. A well-designed elasticity controller helps reducing the cost of hosting services using dynamic resource provisioning and, in the meantime, does not compromise service quality. An elasticity controller often needs to be trained either online or offline in order to make it intelligent enough to make decisions on spawning or removing extra instances when workload increase or decrease. However, there are two main issues on the process of control model training. A significant amount of recent works train the models offline and apply them to an online system. This approach may lead the elasticity controller to make inaccurate decisions since not all parameters can be considered when building the model offline. The complete training of the model consumes large efforts, including modifying system setups and changing system configurations. Worse, some models can even include several dimensions of system parameters. To overcome these limitations, we present the design and evaluation of a self-trained proactive elasticity manager for cloud-based elastic key-value stores. Our elasticity controller uses online profiling and support vector machines (SVM) to provide a black-box performance model of an application’s SLO violation for a given resource demand. The model is dynamically updated to adapt to operating environment changes such as workload pattern variations, data rebalance, changes in data size, etc. We have implemented and evaluated our controller using the Apache Cassandra key-value store in an OpenStack Cloud environment. Our experiments with artificial workload traces shows that our controller guarantees a high level of SLO commitments while keeping the overall resource utilization optimal.

On the area of microcontrollers, a firmware is traditionally built for a very specific hardware configuration. Without special design, there is little chance that the same firmware will run on several hardware platforms with a different sets of peripherals. But running the same firmware on different hardware configurations could have some benefits. It could allow a manufacturer or a sensor network manager to deploy the exact same firmware on all its nodes regardless of their hardware. It would greatly simplify the firmware management, and thus the update process. We know that such a system is possible on larger architectures, such as x86 or even ARM, but in this thesis we target smaller architectures. The typical target here is a sensor network node, running on a very low-power microcontroller. No generic system currently exists to allow a firmware to run on several hardware configurations of this type. In this thesis we present a new generic hardware description system that specifically targets small devices. This system can be integrated with existing frameworks or operating systems for embedded systems so that the firmware can adapt to the hardware it is running on. We show that it is possible by presenting a demonstration prototype using our hardware description system.

This work introduces GeoD and VersionD. GeoD is a causally-consistent georeplication shim for key-value stores. GeoD enables the separation of concerns regarding replication and convergence from key-value stores, while preserving all of their read-only functionality. VersionD extends the feature set of GeoD by providing a versioning API, allowing applications to store multiple values for a given object, and to use context-specific conflict-resolution rules. We also discuss an efficient architecture for optimistically replicated systems, where all the storage replicas persist data in a common shared DHT. Since most key-value stores don’t support versioning and GeoD relies on versioning support from the key-value store, we developed a shim layer that also augments the traditional core API with extra versioning operations and logic.

Measurement of available path capacity with high accuracy over high-speed linksdeployed in cloud and transport backbones is vital for Internet Service Providers(ISP) and for various network applications to know how congested the different linksare in a network and also to do traffic engineering efficiently. State-of-the-art activeprobing method BART (Bandwidth Available in Real-Time) measures the availablepath capacity by injecting the timestamped measurement packets into the network.Active probing measurement technique deeply relies on the ability to generate probepackets at required rate and time-stamped with high precision. While dealing withhigh-speed links (10+ Gbps) links, it is a challenging task for the measurementsystem to generate and time stamp the probe packets within few nano-seconds.This thesis focus on finding a suitable architecture and an optimized algorithmto send and receive measurement packets over high-speed links. Five differentarchitectures: Native UDP sockets, LibPcap, Loadable Kernel Modules, Netmap andData Plane Development Kit (DPDK) are investigated for fast packets processing inLinux systems. Also, the optimized algorithm(s) that suits the specific architecturefor sending and receiving measurement packets at various rates are designed. Theperformance metrics obtained from the measurement systems tested against variousrates and burst size using different architectures are compared and the observationsare discussed and analysed in detail to find the suitable architecture. / Mätning av tillgänglig kapacitet i datanät med hög noggrannhet överhöghastighetslänkar i datacenter och transportnät är grundläggande för attISP:er och olika nätapplikationer ska kunna veta hur överbelastade de olika länkarnaär och för att kunna planera trafiköden på ett effektivt sätt. Den senastetekniken för aktiva mätningar, BART (Bandwidth Available in Real Time), mäterden tillgängliga kapaciteten hos en nätverksväg genom att sända tidsstämplademätpaket genom nätet. Aktiv mätning är helt beroende av möjligheten att kunnagenerera mätpaket med önskad hastighet och att kunna tidsstämpla paketen medhög precision. Vid mätning över höghastighetslänkar (10+ Gbps) är den tekniskautmaningen att generera och tidsstämpla mätpaketen med några få nanosekundersnoggrannhet.Denna avhandling fokuserar på att hitta en passande arkitektur och en optimalalgoritm för att skicka och ta emot mätpaket över höghastighetslänkar. Femolika arkitekturer: rena UDP-socketar, LibPcap, laddbara kernel-moduler, Netmap,och Data Plane Development Kit (DPDK) har testats med avseende på snabbpakethantering i ett Linux-baserat system. Dessutom har optimerade algoritmerdesignats för att kunna skicka och ta emot mätpaket med olika hastigheter för devalda arkitekturerna. Prestanda för de olika mätsystemen har jämförts för olikahastigheter och för olika antal mätpaket i pakettågen och resultaten har diskuteratsoch analyserats i detalj för att hitta en lämplig arkitektur.

Middleboxes are essential to the functioning of today's internet. They are for instance used to secure networks, to enhance performances (e.g., throughput, scalability or end-user latency) or to monitor traffic. Although middleboxes are usually deployed through expensive dedicated hardware, the past 15 years has seen the emergence of a new paradigm: network function virtualisation (NFV). In the NFV context, middleboxes are implemented in software on commodity hardware, thus reducing costs and increasing flexibility. Some of the most recent work even boast performances that are equals to those of hardware middleboxes (e.g., line-rate throughput). However, none of the frameworks that we could find were suited to implement chains of virtualised middleboxes. Indeed, it is often the case that a packet must cross numerous middleboxes when traversing the internet. In order for an NFV deployment to scale out and reduce the network overhead, one would wish to be able to deploy all these middleboxes on the same physical machine. We provide an evaluation of a state-of-the-art NFV framework that shows that the throughput of a chain of 8 middleboxes running on the same server can be as much as 5 times smaller than the throughput of a single middlebox. We then introduce Hyper-NF, a new NFV framework specifically designed for implementing chains of virtualised middleboxes. Hyper-NF eliminates redundant packet and I/O operations. Given a chain of middleboxes, it uses graph search and set theory to generate a single equivalent middlebox that only uses one read and one write operation per packet. Experimentation with middlebox deployments inspired from real-world use cases shows that Hyper-NF achieve constant throughput and latency despite the increasing number of chained middleboxes. Thus, it achieves considerably better performances than traditional deployments. On a chain of 8 virtualised middleboxes, Hyper-NF has a 5 times higher throughput, a 10 times lower latency and uses 8.5 times less CPU cycles per packet. / Middleboxes är absolut nödvändiga i internet idag. De behövs för att säkra nätverken, för att förbättra prestanda (till exempel genomströmning, fördröjning eller skalbarhet) eller för att övervaka nätverkstraffik. Middleboxes brukar vara byggd med dyrt och dedikerad hårdvara men ett nytt paradigm utvecklades in senaste åren: virtualisering av nätverksfunktioner (en: Network Function Virtualization - NFV). I NFV, middleboxes skaffas av flexibel mjukvara på billig råvara hårdvara. Några aktuella ramar även ger samma prestanda som hårdvara middleboxes (t.ex. genomströmning med linjehastighet). Denna ramar är dock inte lämpade för att genomföra kedjor av virtualiserade middleboxes. Ett nätverkspaket måste ofta gå igenom många middleboxes när det korsar internet. Därför skulle det vara bättre att utplacera flera virtuella middleboxes på samma maskin. Så skulle man minska belastningen på nätverket och NFV-utplacering skulle kunna skala ut. Vi visar att även med en state of the art NFV-ram, genomströmningen av en kedjor av 8 middleboxes som körs på samma dator är 5 gånger mindre än den av en enda virtuell middlebox. Vi introducerar Hyper-NF, en ny NFV-ram som vi speciellt utformade för att genomföra kedjor av virtualiserade middleboxes. Hyper-NF eliminerar redundanta I/O och pakets verksamhet. Det förvandlar en kedja av middleboxes till en enda middlebox med liknande funktionalitet genom mängdteori och graf sökning. Den producerade middlebox använder bara ett läs och ett skriv operation per paket. Vi testade Hyper-NF med middlebox-utplacering som inspirerades av verkliga användningsfall. Hyper-NF uppnår konstant prestanda (genomströmning och fördröjning) även om vi öka antalet middleboxes i kedjan. Det är alltså betydligt bättre än traditionella ramar. På en kedja av 8 virtualiserade middleboxes har Hyper-NF en 5 gånger högre genomströmning, en 10 gånger lägre latens och använder 8,5 gånger mindre CPU-cykler per paket.

När ett datorprogram kraschar kan en så kallad kraschdump skapas sominnehåller information över hur arbetsminnet och delar av processorn såg ut vidkraschen [1]. Dessa dumpar är väsentliga för att sedan fixa buggar som orsakarkraschen i programmet.

Detta är en förstudie som undersöker hur man kan utveckla ett systemför funktionsnedsatta och äldre. Systemet är ett hjälpmedelsystem som underlättar för användaren genom att identifiera besökare och låsa upp dörrenfjärrstyrt. Systemet består av kamera, mikrofon och ringklocka på utsidanav ytterdörren. Användaren styr ett elektroniskt dörrlås med hjälp av enmobilapp, där man kan prata med och se besökaren.Vi studerar hur andra system är designade och vilka krav som ställs på dennatypen av system, när de ska vara anpassade till äldre och funktionsnedsatta.Detta görs med hjälp en litteraturstudie och en intervju med en boendekonsulentpå KommSyn Stockholm.För att utveckla ett nytt system måste man ha en bestämd produktutvecklingscykel.Det studeras en ISO standard inom området samt en intervjuarrangeras med en systemutvecklare på Ericsson för att få respons och insiktav näringslivet.Då vårt system använder en kamera påverkas det av kameraövervakningslagen.Studien undersöker hur produkten måste utvecklas och anpassas föratt vara laglig.Vårt system förmedlar trådlöst realtids -ljud och -bild. De olika tekniskalösningar som krävs undersöks och används som underlag till en ny designför en potentiell prototyp. Lösningen använder sig främst av SIP (SessionInitiation Protocol) som är en standard för ip telefoni. / This is a study in which we investigate the design and development of asystem for disabled and elderly. The system is a help system that facilitatesthe user by remotely identifying visitors and unlocking the door. The systemcontains a camera, a microphone and doorbell. The user controls an electronicdoor lock with a smartphone app, where you can talk with and seethe visitor.We study how other systems are designed and which requirements that arenecessary for systems customized for the disabled and elderly. This is donewith a literature study and an interview with a housing consultant at Komm-Syn Stockholm.To develop a new system you need a _xed development cycle. A ISO standardconcerning this is studied and an interview is held with a system developerat Ericsson to get response and insight by the business community.Since our system uses a camera it is a_ected by the CCTV Act. The studyexamines how the product has to be developed and customized to be legal.Our system wirelessly streams real time sound and picture. The technicalsolutions that are required are studied and used in a new design for a potentialprototype. The solution primarily use SIP(Session Initiation Protocol) whichis a standard for ip-telephony.

No description available.