Resource scaling improves the capability of a datacenter or group of datacenters collaborated together to provide resources at low cost in order to meet the demands and objectives of application services, but, it is substantial to determine the requirements of the user, especially in the large projects like XIFI. It is important to estimate the number of users, their arrival rate and types of applications that are often requested for resource allocation, to expand the resource dimensions to proportionate degree. In this study we frame a structure that provides deep insights to comprehend XIFI infrastructure. Furthermore, we model behavior of users that approach the node for resource allocation to run their applications. We aim to provide an understanding on how the user behavior influences the resource scaling in XIFI node. The main objective of this thesis is to investigate different types of applications chosen by users who request for resource allocations and impact of their choice on the resource availability. In the systematic review, a number of deliverables of XIFI to understand the specifications of XIFI architecture are reviewed and analyzed. A model that meets basic requirements, which can be denoted as a XIFI node is developed and the developed design is implemented in a simulator. We simulated the designed structure for 30 iterations and analyzed 10,000 user requests for two cases where total RAM of the node is increased in the second case when compared to the first case. We analyze the reason for the failure of the number of requests and different types of virtual machines for different types of applications, due to unavailable resources. From the obtained results, we conclude that, by increasing total RAM in a XIFI node the failure of average number of requests can be reduced. Also the failure percentage of virtual machines that are to be instantiated, as requested by users decreases when the RAM is scaled to twice the present value. We also conclude that the user behavior that imposes load on the system, decides the degree of scalability of resources in the XIFI node.

The increasingly diverse nature of modern applications makes it critical for future systems to have dynamic resource scaling capabilities which enable them to adapt their resource usage to meet user requirements. Such mechanisms should be both fine-grained in nature for resource-efficient operation and also provide a high scaling range to support a variety of applications with diverse needs. To this end, heterogeneous platforms, consisting of components with varying characteristics, have been proposed to provide improved performance/efficiency than homogeneous configurations, by making it possible to execute applications on the most suitable component. However, introduction of such heterogeneous architectural components requires system software to embrace complexity associated with heterogeneity for managing them efficiently. Diversity across vendors and rapidly changing hardware make it difficult to incorporate heterogeneity-aware resource management mechanisms into mainstream systems, affecting the widespread adoption of these platforms. Addressing these issues, this dissertation presents novel abstractions and mechanisms for heterogeneous platforms which decouple heterogeneity from management operations by masking the differences due to heterogeneity from applications. By exporting a homogeneous interface over heterogeneous components, it proposes the scalable 'resource state' abstraction, allowing applications to express their resource requirements which then are dynamically and transparently mapped to heterogeneous resources underneath. The proposed approach is explored for both modern mobile devices where power is a key resource and for cloud computing environments where platform resource usage has monetary implications, resulting in HeteroMates and HeteroVisor solutions. In addition, it also highlights the need for hardware and system software to consider multiple resources together to obtain desirable gains from such scaling mechanisms. The solutions presented in this dissertation open ways for utilizing future heterogeneous platforms to provide on-demand performance, as well as resource-efficient operation, without disrupting the existing software stack.

The pervasive availability of streaming data from various sources is driving todays’ enterprises to acquire low-latency big data streaming applications (BDSAs) for extracting useful information. In parallel, recent advances in technology have made it easier to collect, process and store these data streams in the cloud. For most enterprises, gaining insights from big data is immensely important for maintaining competitive advantage. However, majority of enterprises have difficulty managing the multitude of BDSAs and the complex issues cloud technologies present, giving rise to the incorporation of cloud service brokers (CSBs). Generally, the main objective of the CSB is to maintain the heterogeneous quality of service (QoS) of BDSAs while minimizing costs. To achieve this goal, the cloud, although with many desirable features, exhibits major challenges — resource prediction and resource allocation — for CSBs. First, most stream processing systems allocate a fixed amount of resources at runtime, which can lead to under- or over-provisioning as BDSA demands vary over time. Thus, obtaining optimal trade-off between QoS violation and cost requires accurate demand prediction methodology to prevent waste, degradation or shutdown of processing. Second, coordinating resource allocation and pricing decisions for self-interested BDSAs to achieve fairness and efficiency can be complex. This complexity is exacerbated with the recent introduction of containers. This dissertation addresses the cloud resource elasticity management issues for CSBs as follows: First, we provide two contributions to the resource prediction challenge; we propose a novel layered multi-dimensional hidden Markov model (LMD-HMM) framework for managing time-bounded BDSAs and a layered multi-dimensional hidden semi-Markov model (LMD-HSMM) to address unbounded BDSAs. Second, we present a container resource allocation mechanism (CRAM) for optimal workload distribution to meet the real-time demands of competing containerized BDSAs. We formulate the problem as an n-player non-cooperative game among a set of heterogeneous containerized BDSAs. Finally, we incorporate a dynamic incentive-compatible pricing scheme that coordinates the decisions of self-interested BDSAs to maximize the CSB’s surplus. Experimental results demonstrate the effectiveness of our approaches.

The user experience offered by a video streaming service plays a fundamental role in customer satisfaction. This experience can be degraded by poor playback quality and buffering issues. These problems can be caused by a user demand that is higher than the video streaming service capacity. Resource scaling methods can increase the available resources to cover the need. However, most resource scaling systems are reactive and scale up in an automated fashion when a certain demand threshold is exceeded. During popular live streaming content, the demand can be so high that even by scaling up at the last minute, the system might still be momentarily under-provisioned, resulting in a bad user experience. The solution to this problem is proactive scaling which is event-based, using content-related information to scale up or down, according to knowledge from past events. As a result, proactive resource scaling is a key factor in providing reliable video streaming services. Users viewing habits heavily affect demand. To provide an accurate model for proactive resource scaling tools, these habits need to be modelled. This thesis provides such a forecasting model for user views that can be used by a proactive resource scaling mechanism. This model is created by applying machine learning algorithms to data from both live TV and over-the-top streaming services. To produce a model with satisfactory accuracy, numerous data attributes were considered relating to users, content and content providers. The findings of this thesis show that user viewing demand can be modelled with high accuracy, without heavily relying on user-related attributes but instead by analysing past event logs and with knowledge of the schedule of the content provider, whether it is live tv or a video streaming service. / Användarupplevelsen som erbjuds av en videostreamingtjänst spelar en grundläggande roll för kundnöjdheten. Denna upplevelse kan försämras av dålig uppspelningskvalitet och buffertproblem. Dessa problem kan orsakas av en efterfrågan från användare som är högre än videostreamingtjänstens kapacitet. Resursskalningsmetoder kan öka tillgängliga resurser för att täcka behovet. De flesta resursskalningssystem är dock reaktiva och uppskalas automatiskt när en viss behovströskel överskrids. Under populärt livestreaminginnehåll kan efterfrågan vara så hög att även genom att skala upp i sista minuten kan systemet fortfarande vara underutnyttjat tillfälligt, vilket resulterar i en dålig användarupplevelse. Lösningen på detta problem är proaktiv skalning som är händelsebaserad och använder innehållsrelaterad information för att skala upp eller ner, enligt kunskap från tidigare händelser. Som ett resultat är proaktiv resursskalning en nyckelfaktor för att tillhandahålla tillförlitliga videostreamingtjänster. Användares visningsvanor påverkar efterfrågan kraftigt. För att ge en exakt modell för proaktiva resursskalningsverktyg måste dessa vanor modelleras. Denna avhandling ger en sådan prognosmodell för användarvyer som kan användas av en proaktiv resursskalningsmekanism. Denna modell är skapad genom att använda maskininlärningsalgoritmer på data från både live-TV och streamingtjänster. För att producera en modell med tillfredsställande noggrannhet ansågs ett flertal dataattribut relaterade till användare, innehåll och innehållsleverantörer. Resultaten av den här avhandlingen visar att efterfrågan på användare kan modelleras med hög noggrannhet utan att starkt förlita sig på användarrelaterade attribut utan istället genom att analysera tidigare händelseloggar och med kunskap om innehållsleverantörens schema, vare sig det är live-tv eller tjänster för videostreaming.