Würzburger Beiträge zur Leistungsbewertung Verteilter Systeme
Filtern
Volltext vorhanden
- ja (35)
Gehört zur Bibliographie
- ja (35)
Erscheinungsjahr
Dokumenttyp
- Dissertation (35)
Schlagworte
- Leistungsbewertung (25)
- Quality of Experience (9)
- Netzwerk (8)
- Performance Evaluation (7)
- Optimierung (6)
- Software Defined Networking (5)
- Dienstgüte (4)
- Optimization (4)
- Overlay-Netz (4)
- Routing (4)
Institut
02/08
Mobile telecommunication systems of the 3.5th generation (3.5G) constitute a first step towards the requirements of an all-IP world. As the denotation suggests, 3.5G systems are not completely new designed from scratch. Instead, they are evolved from existing 3G systems like UMTS or cdma2000. 3.5G systems are primarily designed and optimized for packet-switched best-effort traffic, but they are also intended to increase system capacity by exploiting available radio resources more efficiently. Systems based on cdma2000 are enhanced with 1xEV-DO (EV-DO: evolution, data-optimized). In the UMTS domain, the 3G partnership project (3GPP) specified the High Speed Packet Access (HSPA) family, consisting of High Speed Downlink Packet Access (HSDPA) and its counterpart High Speed Uplink Packet Access (HSUPA) or Enhanced Uplink. The focus of this monograph is on HSPA systems, although the operation principles of other 3.5G systems are similar. One of the main contributions of our work are performance models which allow a holistic view on the system. The models consider user traffic on flow-level, such that only on significant changes of the system state a recalculation of parameters like bandwidth is necessary. The impact of lower layers is captured by stochastic models. This approach combines accurate modeling and the ability to cope with computational complexity. Adopting this approach to HSDPA, we develop a new physical layer abstraction model that takes radio resources, scheduling discipline, radio propagation and mobile device capabilities into account. Together with models for the calculation of network-wide interference and transmit powers, a discrete-event simulation and an analytical model based on a queuing-theoretical approach are proposed. For the Enhanced Uplink, we develop analytical models considering independent and correlated other-cell interference.
03/08
The Internet sees an ongoing transformation process from a single best-effort service network into a multi-service network. In addition to traditional applications like e-mail,WWW-traffic, or file transfer, future generation networks (FGNs) will carry services with real-time constraints and stringent availability and reliability requirements like Voice over IP (VoIP), video conferencing, virtual private networks (VPNs) for finance, other real-time business applications, tele-medicine, or tele-robotics. Hence, quality of service (QoS) guarantees and resilience to failures are crucial characteristics of an FGN architecture. At the same time, network operations must be efficient. This necessitates sophisticated mechanisms for the provisioning and the control of future communication infrastructures. In this work we investigate such echanisms for resilient FGNs. There are many aspects of the provisioning and control of resilient FGNs such as traffic matrix estimation, traffic characterization, traffic forecasting, mechanisms for QoS enforcement also during failure cases, resilient routing, or calability concerns for future routing and addressing mechanisms. In this work we focus on three important aspects for which performance analysis can deliver substantial insights: load balancing for multipath Internet routing, fast resilience concepts, and advanced dimensioning techniques for resilient networks. Routing in modern communication networks is often based on multipath structures, e.g., equal-cost multipath routing (ECMP) in IP networks, to facilitate traffic engineering and resiliency. When multipath routing is applied, load balancing algorithms distribute the traffic over available paths towards the destination according to pre-configured distribution values. State-of-the-art load balancing algorithms operate either on the packet or the flow level. Packet level mechanisms achieve highly accurate traffic distributions, but are known to have negative effects on the performance of transport protocols and should not be applied. Flow level mechanisms avoid performance degradations, but at the expense of reduced accuracy. These inaccuracies may have unpredictable effects on link capacity requirements and complicate resource management. Thus, it is important to exactly understand the accuracy and dynamics of load balancing algorithms in order to be able to exercise better network control. Knowing about their weaknesses, it is also important to look for alternatives and to assess their applicability in different networking scenarios. This is the first aspect of this work. Component failures are inevitable during the operation of communication networks and lead to routing disruptions if no special precautions are taken. In case of a failure, the robust shortest-path routing of the Internet reconverges after some time to a state where all nodes are again reachable – provided physical connectivity still exists. But stringent availability and reliability criteria of new services make a fast reaction to failures obligatory for resilient FGNs. This led to the development of fast reroute (FRR) concepts for MPLS and IP routing. The operations of MPLS-FRR have already been standardized. Still, the standards leave some degrees of freedom for the resilient path layout and it is important to understand the tradeoffs between different options for the path layout to efficiently provision resilient FGNs. In contrast, the standardization for IP-FRR is an ongoing process. The applicability and possible combinations of different concepts still are open issues. IP-FRR also facilitates a comprehensive resilience framework for IP routing covering all steps of the failure recovery cycle. These points constitute another aspect of this work. Finally, communication networks are usually over-provisioned, i.e., they have much more capacity installed than actually required during normal operation. This is a precaution for various challenges such as network element failures. An alternative to this capacity overprovisioning (CO) approach is admission control (AC). AC blocks new flows in case of imminent overload due to unanticipated events to protect the QoS for already admitted flows. On the one hand, CO is generally viewed as a simple mechanism, AC as a more complex mechanism that complicates the network control plane and raises interoperability issues. On the other hand, AC appears more cost-efficient than CO. To obtain advanced provisioning methods for resilient FGNs, it is important to find suitable models for irregular events, such as failures and different sources of overload, and to incorporate them into capacity dimensioning methods. This allows for a fair comparison between CO and AC in various situations and yields a better understanding of the strengths and weaknesses of both concepts. Such an advanced capacity dimensioning method for resilient FGNs represents the third aspect of this work.
01/08
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.
01/07
Performance Evaluation of Efficient Resource Management Concepts for Next Generation IP Networks
(2007)
Next generation networks (NGNs) must integrate the services of current circuit-switched telephone networks and packet-switched data networks. This convergence towards a unified communication infrastructure necessitates from the high capital expenditures (CAPEX) and operational expenditures (OPEX) due to the coexistence of separate networks for voice and data. In the end, NGNs must offer the same services as these legacy networks and, therefore, they must provide a low-cost packet-switched solution with real-time transport capabilities for telephony and multimedia applications. In addition, NGNs must be fault-tolerant to guarantee user satisfaction and to support business-critical processes also in case of network failures. A key technology for the operation of NGNs is the Internet Protocol (IP) which evolved to a common and well accepted standard for networking in the Internet during the last 25 years. There are two basically different approaches to achieve QoS in IP networks. With capacity overprovisioning (CO), an IP network is equipped with sufficient bandwidth such that network congestion becomes very unlikely and QoS is maintained most of the time. The second option to achieve QoS in IP networks is admission control (AC). AC represents a network-inherent intelligence that admits real-time traffic flows to a single link or an entire network only if enough resources are available such that the requirements on packet loss and delay can be met. Otherwise, the request of a new flow is blocked. This work focuses on resource management and control mechanisms for NGNs, in particular on AC and associated bandwidth allocation methods. The first contribution consists of a new link-oriented AC method called experience-based admission control (EBAC) which is a hybrid approach dealing with the problems inherent to conventional AC mechanisms like parameter-based or measurement-based AC (PBAC/MBAC). PBAC provides good QoS but suffers from poor resource utilization and, vice versa, MBAC uses resources efficiently but is susceptible to QoS violations. Hence, EBAC aims at increasing the resource efficiency while maintaining the QoS which increases the revenues of ISPs and postpones their CAPEX for infrastructure upgrades. To show the advantages of EBAC, we first review today’s AC approaches and then develop the concept of EBAC. EBAC is a simple mechanism that safely overbooks the capacity of a single link to increase its resource utilization. We evaluate the performance of EBAC by its simulation under various traffic conditions. The second contribution concerns dynamic resource allocation in transport networks which implement a specific network admission control (NAC) architecture. In general, the performance of different NAC systems may be evaluated by conventional methods such as call blocking analysis which has often been applied in the context of multi-service asynchronous transfer mode (ATM) networks. However, to yield more practical results than abstract blocking probabilities, we propose a new method to compare different AC approaches by their respective bandwidth requirements. To present our new method for comparing different AC systems, we first give an overview of network resource management (NRM) in general. Then we present the concept of adaptive bandwidth allocation (ABA) in capacity tunnels and illustrate the analytical performance evaluation framework to compare different AC systems by their capacity requirements. Different network characteristics influence the performance of ABA. Therefore, the impact of various traffic demand models and tunnel implementations, and the influence of resilience requirements is investigated. In conclusion, the resources in NGNs must be exclusively dedicated to admitted traffic to guarantee QoS. For that purpose, robust and efficient concepts for NRM are required to control the requested bandwidth with regard to the available transmission capacity. Sophisticated AC will be a key function for NRM in NGNs and, therefore, efficient resource management concepts like experience-based admission control and adaptive bandwidth allocation for admission-controlled capacity tunnels, as presented in this work are appealing for NGN solutions.
03/04
This work is subdivided into two main areas: resilient admission control and resilient routing. The work gives an overview of the state of the art of quality of service mechanisms in communication networks and proposes a categorization of admission control (AC) methods. These approaches are investigated regarding performance, more precisely, regarding the potential resource utilization by dimensioning the capacity for a network with a given topology, traffic matrix, and a required flow blocking probability. In case of a failure, the affected traffic is rerouted over backup paths which increases the traffic rate on the respective links. To guarantee the effectiveness of admission control also in failure scenarios, the increased traffic rate must be taken into account for capacity dimensioning and leads to resilient AC. Capacity dimensioning is not feasible for existing networks with already given link capacities. For the application of resilient NAC in this case, the size of distributed AC budgets must be adapted according to the traffic matrix in such a way that the maximum blocking probability for all flows is minimized and that the capacity of all links is not exceeded by the admissible traffic rate in any failure scenario. Several algorithms for the solution of that problem are presented and compared regarding their efficiency and fairness. A prototype for resilient AC was implemented in the laboratories of Siemens AG in Munich within the scope of the project KING. Resilience requires additional capacity on the backup paths for failure scenarios. The amount of this backup capacity depends on the routing and can be minimized by routing optimization. New protection switching mechanisms are presented that deviate the traffic quickly around outage locations. They are simple and can be implemented, e.g, by MPLS technology. The Self-Protecting Multi-Path (SPM) is a multi-path consisting of disjoint partial paths. The traffic is distributed over all faultless partial paths according to an optimized load balancing function both in the working case and in failure scenarios. Performance studies show that the network topology and the traffic matrix also influence the amount of required backup capacity significantly. The example of the COST-239 network illustrates that conventional shortest path routing may need 50% more capacity than the optimized SPM if all single link and node failures are protected.