Stefan Köhler

MPLS traffic engineering in OSPF networks—A combined approach

One of the main reasons for the development of MPLS was the need for flexible Traffic Engineering (TE) in IP networks ([8], [1]). Additionally a lot of IP service providers have found that TE of their networks can have a tactical and strategic value [2]. The first part of the paper introduces a linear optimization approach for TE in MPLS networks. Most remarkable on the presented approach in comparison to other linear approaches is the additive objective function, which optimizes both the maximum utilization as well as the average utilization. In practice, most of the existing OSPF or IS-IS networks will have a transition phase between the pure IGP routing and the MPLS driven approach. Thus, we extend our optimization approach from a regular IGP network to a mixed IGP-MPLS environment. To avoid configuration complexity and state space explosion in MPLS devices, it could also make sense to start with a MPLS environment and take advantage of the configuration simplicity and the state space reduction capability of IGP protocols. In addition the paper includes a performance comparison between default configured OSPF, weight optimized OSPF and pure MPLS networks and shows the benefit of OSPF optimization and MPLS TE. Based on this comparison we can offer some valuable clues to decide in favor of a local or global TE system.

The dependence of Internet user traffic characteristics on access speed

A variety of Internet traffic characterization and source models already exist. Since the origin of the analyzed data differs, a comparison of the reported traffic characteristics is difficult. In this paper we analyze an IP packet trace and relate the trace data to the user access speed recorded at the same time. The obtained characteristics are contrasted with a second data set measured within an ADSL field trial. This second data set represents a limit in Internet usage, since, the high speed access was provided free of charge. This comparison leads to some interesting insights showing whether and how the access speed influences the Internet user behavior and thus the generated source traffic. The derived Internet traffic characterization is a first step to a generalized parametric Internet traffic model.

Performance Comparison of Different Class-and-Drop Treatment of Data and Acknowledgements in DiffServ IP Networks

In differentiated services IP networks the sender of a TCP connection determines the class of data packets he emits. The receiver chooses the class of the acknowledgements sent back, independently of the received class. In this work, we examine the impact of different drop precedence within a class and the assignment of different classes for data and acknowledgements. The results show that the throughput of a TCP connection depends not only on the data class, but also on the right choice for the acknowledgements. Some combinations of classes for data and ACKs could even lead to an “unfair” use of bandwidth. On the other hand, for a high throughput the selection of the drop precedence is in most cases only important for data packets.

First Description of Reduced Pyruvate Dehydrogenase Enzyme Activity Following Subarachnoid Hemorrhage (SAH)

Object: Several previous studies reported metabolic derangements and an accumulation of metabolic products in the early phase of experimental subarachnoid hemorrhage (SAH), which may contribute to secondary brain damage. This may be a result of deranged oxygen utilization due to enzymatic dysfunction in aerobic glucose metabolism. This study was performed to investigate, if pyruvate dehydrogenase enzyme (PDH) is affected in its activity giving further hints for a derangement of oxidative metabolism. Methods: Eighteen male Sprague-Dawley rats were randomly assigned to one of two experimental groups (n = 9): (1) SAH induced by the endovascular filament model and (2) sham-operated controls. Mean arterial blood pressure (MABP), intracranial pressure (ICP), and local cerebral blood flow (LCBF; laser-Doppler flowmetry) were continuously monitored from 30 min before until 3 h after SAH. Thereafter, the animals were sacrificed and PDH activity was measured by ELISA. Results: PDH activity was significantly reduced in animals subjected to SAH compared to controls. Conclusion: The results of this study demonstrate for the first time a reduction of PDH activity following SAH, independent of supply of substrates and may be an independent factor contributing to a derangement of oxidative metabolism, failure of oxygen utilization, and secondary brain damage.