Refine
Has Fulltext
- yes (29)
Is part of the Bibliography
- yes (29)
Year of publication
Document Type
- Doctoral Thesis (29)
Language
- English (29)
Keywords
- Leistungsbewertung (29) (remove)
Institute
Overlay networks establish logical connections between users on top of the physical network. While randomly connected overlay networks provide only a best effort service, a new generation of structured overlay systems based on Distributed Hash Tables (DHTs) was proposed by the research community. However, there is still a lack of understanding the performance of such DHTs. Additionally, those architectures are highly distributed and therefore appear as a black box to the operator. Yet an operator does not want to lose control over his system and needs to be able to continuously observe and examine its current state at runtime. This work addresses both problems and shows how the solutions can be combined into a more self-organizing overlay concept. At first, we evaluate the performance of structured overlay networks under different aspects and thereby illuminate in how far such architectures are able to support carrier-grade applications. Secondly, to enable operators to monitor and understand their deployed system in more detail, we introduce both active as well as passive methods to gather information about the current state of the overlay network.
Content Delivery Networks (CDNs) are networks that distribute content in the Internet. CDNs are increasingly responsible for the largest share of traffic in the Internet. CDNs distribute popular content to caches in many geographical areas to save bandwidth by avoiding unnecessary multihop retransmission. By bringing the content geographically closer to the user, CDNs also reduce the latency of the services.
Besides end users and content providers, which require high availability of high quality content, CDN providers and Internet Service Providers (ISPs) are interested in an efficient operation of CDNs. In order to ensure an efficient replication of the content, CDN providers have a network of (globally) distributed interconnected datacenters at different points of presence (PoPs). ISPs aim to provide reliable and high speed Internet access. They try to keep the load on the network low and to reduce cost for connectivity with other ISPs.
The increasing number of mobile devices such as smart phones and tablets, high definition video content and high resolution displays result in a continuous growth in mobile traffic. This growth in mobile traffic is further accelerated by newly emerging services, such as mobile live streaming and broadcasting services. The steep increase in mobile traffic is expected to reach by 2018 roughly 60% of total network traffic, the majority of which will be video. To handle the growth in mobile networks, the next generation of 5G mobile networks is designed to have higher access rates and an increased densification of the network infrastructure. With the explosion of access rates and number of base stations the backhaul of wireless networks will become congested.
To reduce the load on the backhaul, the research community suggests installing local caches in gateway routers between the wireless network and the Internet, in base stations of different sizes, and in end-user devices. The local deployment of caches allows keeping the traffic within the ISPs network. The caches are organized in a hierarchy, where caches in the lowest tier are requested first. The request is forwarded to the next tier, if the requested object is not found. Appropriate evaluation methods are required to optimally dimension the caches dependent on the traffic characteristics and the available resources. Additionally methods are necessary that allow performance evaluation of backhaul bandwidth aggregation systems, which further reduce the load on the backhaul.
This thesis analyses CDNs utilizing locally available resources and develops the following evaluations and optimization approaches: Characterization of CDNs and distribution of resources in the Internet, analysis and optimization of hierarchical caching systems with bandwidth constraints and performance evaluation of bandwidth aggregation systems.
In this thesis various aspects of Quality of Experience (QoE) research are examined. The work is divided into three major blocks: QoE Assessment, QoE Monitoring, and VNF Performance Evaluation. First, prominent cloud applications such as Google Docs and a cloud-based photo album are explored. The QoE is characterized and the influence of packet loss and delay is studied. Afterwards, objective QoE monitoring for HTTP Adaptive Video Streaming (HAS) in the cloud is investigated. Additionally, by using a Virtual Network Function (VNF) for QoE monitoring in the cloud, the feasibility of an interworking of Network Function Virtualization (NFV) and cloud paradigm is evaluated. To this end, a VNF that exploits deep packet inspection technique was used to parse the video traffic. An algorithm is then designed accordingly to estimate video quality and QoE based on network and application layer parameters. To assess the accuracy of the estimation, the VNF is measured in different scenarios under different network QoS and the virtual environment of the cloud architecture. The insights show that the different geographical deployments of the VNF influence the accuracy of the video quality and QoE estimation. Various Service Function Chain (SFC) placement algorithms have been proposed and compared in the context of edge cloud networks. On the one hand, this research is aimed at cloud service providers by providing methods for evaluating QoE for cloud applications. On the other hand, network operators can learn the pitfalls and disadvantages of using the NFV paradigm for such a QoE monitoring mechanism.
Serverless computing is an emerging cloud computing paradigm that offers a highlevel
application programming model with utilization-based billing. It enables the
deployment of cloud applications without managing the underlying resources or
worrying about other operational aspects. Function-as-a-Service (FaaS) platforms
implement serverless computing by allowing developers to execute code on-demand
in response to events with continuous scaling while having to pay only for the
time used with sub-second metering. Cloud providers have further introduced
many fully managed services for databases, messaging buses, and storage that also
implement a serverless computing model. Applications composed of these fully
managed services and FaaS functions are quickly gaining popularity in both industry
and in academia.
However, due to this rapid adoption, much information surrounding serverless
computing is inconsistent and often outdated as the serverless paradigm evolves.
This makes the performance engineering of serverless applications and platforms
challenging, as there are many open questions, such as: What types of applications
is serverless computing well suited for, and what are its limitations? How should
serverless applications be designed, configured, and implemented? Which design
decisions impact the performance properties of serverless platforms and how can
they be optimized? These and many other open questions can be traced back to an
inconsistent understanding of serverless applications and platforms, which could
present a major roadblock in the adoption of serverless computing.
In this thesis, we address the lack of performance knowledge surrounding serverless
applications and platforms from multiple angles: we conduct empirical studies
to further the understanding of serverless applications and platforms, we introduce
automated optimization methods that simplify the operation of serverless applications,
and we enable the analysis of design tradeoffs of serverless platforms by
extending white-box performance modeling.
This thesis contributes to several issues in the context of SDN and NFV, with an emphasis on performance and management.
The main contributions are guide lines for operators migrating to software-based networks, as well as an analytical model for the packet processing in a Linux system using the Kernel NAPI.
In this doctoral thesis we cover the performance evaluation of next generation data plane architectures, comprised of complex software as well as programmable hardware components that allow fine granular configuration. In the scope of the thesis we propose mechanisms to monitor the performance of singular components and model key performance indicators of software based packet processing solutions. We present novel approaches towards network abstraction that allow the integration of heterogeneous data plane technologies into a singular network while maintaining total transparency between control and data plane. Finally, we investigate a full, complex system consisting of multiple software-based solutions and perform a detailed performance analysis. We employ simulative approaches to investigate overload control mechanisms that allow efficient operation under adversary conditions. The contributions of this work build the foundation for future research in the areas of network softwarization and network function virtualization.
While developing modern applications, it is necessary to ensure an efficient and performant communication between different applications. In current environments, a middleware software is used, which supports the publish/subscribe communication pattern. Using this communication pattern, a publisher sends information encapsulated in messages to the middleware. A subscriber registers its interests at the middleware. The monograph describes three different steps to determine the performance of such a system. In a first step, the message throughput performance of a publish/subscribe in different scenarios is measured using a Java Message Service (JMS) based implementation. In the second step the maximum achievable message throughput is described by adapted models depending on the filter complexity and the replication grade. Using the model, the performance characteristics of a specific system in a given scenario can be determined. These numbers are used for the queuing model described in the third part of the thesis, which supports the dimensioning of a system in realistic scenarios. Additionally, we introduce a method to approximate an M/G/1 system numerically in an efficient way, which can be used for real time analysis to predict the expected performance in a certain scenario. Finally, the analytical model is used to investigate different possibilities to ensure the scalability of the maximum achievable message throughput of the overall system.
Computer systems have replaced human work-force in many parts of everyday life, but there still exists a large number of tasks that cannot be automated, yet. This also includes tasks, which we consider to be rather simple like the categorization of image content or subjective ratings. Traditionally, these tasks have been completed by designated employees or outsourced to specialized companies. However, recently the crowdsourcing paradigm is more and more applied to complete such human-labor intensive tasks. Crowdsourcing aims at leveraging the huge number of Internet users all around the globe, which form a potentially highly available, low-cost, and easy accessible work-force.
To enable the distribution of work on a global scale, new web-based services emerged, so called crowdsourcing platforms, that act as mediator between employers posting tasks and workers completing tasks. However, the crowdsourcing approach, especially the large anonymous worker crowd, results in two types of challenges. On the one hand, there are technical challenges like the dimensioning of crowdsourcing platform infrastructure or the interconnection of crowdsourcing platforms and machine clouds to build hybrid services. On the other hand, there are conceptual challenges like identifying reliable workers or migrating traditional off-line work to the crowdsourcing environment. To tackle these challenges, this monograph analyzes and models current crowdsourcing systems to optimize crowdsourcing workflows and the underlying infrastructure. First, a categorization of crowdsourcing tasks and platforms is developed to derive generalizable properties. Based on this categorization and an exemplary analysis of a commercial crowdsourcing platform, models for different aspects of crowdsourcing platforms and crowdsourcing mechanisms are developed. A special focus is put on quality assurance mechanisms for crowdsourcing tasks, where the models are used to assess the suitability and costs of existing approaches for different types of tasks. Further, a novel quality assurance mechanism solely based on user-interactions is proposed and its feasibility is shown. The findings from the analysis of existing platforms, the derived models, and the developed quality assurance mechanisms are finally used to derive best practices for two crowdsourcing use-cases, crowdsourcing-based network measurements and crowdsourcing-based subjective user studies. These two exemplary use-cases cover aspects typical for a large range of crowdsourcing tasks and illustrated the potential benefits, but also resulting challenges when using crowdsourcing.
With the ongoing digitalization and globalization of the labor markets, the crowdsourcing paradigm is expected to gain even more importance in the next years. This is already evident in the currently new emerging fields of crowdsourcing, like enterprise crowdsourcing or mobile crowdsourcing. The models developed in the monograph enable platform providers to optimize their current systems and employers to optimize their workflows to increase their commercial success. Moreover, the results help to improve the general understanding of crowdsourcing systems, a key for identifying necessary adaptions and future improvements.
Routing is one of the most important issues in any communication network. It defines on which path packets are transmitted from the source of a connection to the destination. It allows to control the distribution of flows between different locations in the network and thereby is a means to influence the load distribution or to reach certain constraints imposed by particular applications. As failures in communication networks appear regularly and cannot be completely avoided, routing is required to be resilient against such outages, i.e., routing still has to be able to forward packets on backup paths even if primary paths are not working any more.
Throughout the years, various routing technologies have been introduced that are very different in their control structure, in their way of working, and in their ability to handle certain failure cases. Each of the different routing approaches opens up their own specific questions regarding configuration, optimization, and inclusion of resilience issues. This monograph investigates, with the example of three particular routing technologies, some concrete issues regarding the analysis and optimization of resilience. It thereby contributes to a better general, technology-independent understanding of these approaches and of their diverse potential for the use in future network architectures.
The first considered routing type, is decentralized intra-domain routing based on administrative IP link costs and the shortest path principle. Typical examples are common today's intra-domain routing protocols OSPF and IS-IS. This type of routing includes automatic restoration abilities in case of failures what makes it in general very robust even in the case of severe network outages including several failed components. Furthermore, special IP-Fast Reroute mechanisms allow for a faster reaction on outages. For routing based on link costs, traffic engineering, e.g. the optimization of the maximum relative link load in the network, can be done indirectly by changing the administrative link costs to adequate values.
The second considered routing type, MPLS-based routing, is based on the a priori configuration of primary and backup paths, so-called Label Switched Paths. The routing layout of MPLS paths offers more freedom compared to IP-based routing as it is not restricted by any shortest path constraints but any paths can be setup. However, this in general involves a higher configuration effort.
Finally, in the third considered routing type, typically centralized routing using a Software Defined Networking (SDN) architecture, simple switches only forward packets according to routing decisions made by centralized controller units. SDN-based routing layouts offer the same freedom as for explicit paths configured using MPLS. In case of a failure, new rules can be setup by the controllers to continue the routing in the reduced topology. However, new resilience issues arise caused by the centralized architecture. If controllers are not reachable anymore, the forwarding rules in the single nodes cannot be adapted anymore. This might render a rerouting in case of connection problems in severe failure scenarios infeasible.
In future telecommunication systems, we observe an increasing diversity of access networks. The separation of transport services and applications or services leads to multi-network services, i.e., a future service has to work transparently to the underlying network infrastructure. Multi-network services with edge-based intelligence, like P2P file sharing or the Skype VoIP service, impose new traffic control paradigms on the future Internet. Such services adapt the amount of consumed bandwidth to reach different goals. A selfish behavior tries to keep the QoE of a single user above a certain level. Skype, for instance, repeats voice samples depending on the perceived end-to-end loss. From the viewpoint of a single user, the replication of voice data overcomes the degradation caused by packet loss and enables to maintain a certain QoE. The cost for this achievement is a higher amount of consumed bandwidth. However, if the packet loss is caused by congestion in the network, this additionally required bandwidth even worsens the network situation. Altruistic behavior, on the other side, would reduce the bandwidth consumption in such a way that the pressure on the network is released and thus the overall network performance is improved. In this monograph, we analyzed the impact of the overlay, P2P, and QoE paradigms in future Internet applications and the interactions from the observing user behavior. The shift of intelligence toward the edge is accompanied by a change in the emerging user behavior and traffic profile, as well as a change from multi-service networks to multi-networks services. In addition, edge-based intelligence may lead to a higher dynamics in the network topology, since the applications are often controlled by an overlay network, which can rapidly change in size and structure as new nodes can leave or join the overlay network in an entirely distributed manner. As a result, we found that the performance evaluation of such services provides new challenges, since novel key performance factors have to be first identified, like pollution of P2P systems, and appropriate models of the emerging user behavior are required, e.g. taking into account user impatience. As common denominator of the presented studies in this work, we focus on a user-centric view when evaluating the performance of future Internet applications. For a subscriber of a certain application or service, the perceived quality expressed as QoE will be the major criterion of the user's satisfaction with the network and service providers. We selected three different case studies and characterized the application's performance from the end user's point of view. Those are (1) cooperation in mobile P2P file sharing networks, (2) modeling of online TV recording services, and (3) QoE of edge-based VoIP applications. The user-centric approach facilitates the development of new mechanisms to overcome problems arising from the changing user behavior. An example is the proposed CycPriM cooperation strategy, which copes with selfish user behavior in mobile P2P file sharing system. An adequate mechanism has also been shown to be efficient in a heterogeneous B3G network with mobile users conducting vertical handovers between different wireless access technologies. The consideration of the user behavior and the user perceived quality guides to an appropriate modeling of future Internet applications. In the case of the online TV recording service, this enables the comparison between different technical realizations of the system, e.g. using server clusters or P2P technology, to properly dimension the installed network elements and to assess the costs for service providers. Technologies like P2P help to overcome phenomena like flash crowds and improve scalability compared to server clusters, which may get overloaded in such situations. Nevertheless, P2P technology invokes additional challenges and different user behavior to that seen in traditional client/server systems. Beside the willingness to share files and the churn of users, peers may be malicious and offer fake contents to disturb the data dissemination. Finally, the understanding and the quantification of QoE with respect to QoS degradations permits designing sophisticated edge-based applications. To this end, we identified and formulated the IQX hypothesis as an exponential interdependency between QoE and QoS parameters, which we validated for different examples. The appropriate modeling of the emerging user behavior taking into account the user's perceived quality and its interactions with the overlay and P2P paradigm will finally help to design future Internet applications.