arXiv subjects

Pascal Anelli

Publications and source records attributed to Pascal Anelli.

5 recordsLinked to original sources

arXiv2017

Poor Peering: a reflexion about a RIXP

Since more than twenty years, Internet evolves as well as users' needs. and the users' needs too. The need of bandwidth growths every day with the new usages. In this context, with megabit per second count. There is a real business behind our free access to web site. From a group of island, the economy of bandwidth can be a major challenge. Installing a Regional Internet eXchange Point (RIXP) can be a solution to keep regional Internet traffic. This paper brings a reflexion about where to install a RIXP in the Indian Ocean Area (IOA). As there is a lot of different countries and stakes, we (stay) only on an scientist approach.

cs.NI↗

arXiv2017

Where is My Next Hop ? The Case of Indian Ocean Islands

Internet has become a foundation of our modern society. However, all regions or countries do not have the same Internet access regarding quality especially in the Indian Ocean Area (IOA). To improve this quality it is important to have a deep knowledge of the Internet physical and logical topology and associated performance. However, these knowledges are not shared by Internet service providers. In this paper, we describe a large scale measurement study in which we deploy probes in different IOA countries, we generate network traces, develop a tool to extract useful information and analyze these information. We show that most of the IOA traffic exits through one point even if there exists multiple exit points.

cs.NI↗

arXiv2017

How long delays impact TCP performance for a connectivity from Reunion Island ?

TCP is the protocol of transport the most used in the Internet and have a heavy-dependence on delay. Reunion Island have a specific Internet connection, based on main links to France, located 10.000 km away. As a result, the minimal delay between Reunion Island and France is around 180 ms. In this paper, we will study TCP traces collected in Reunion Island University. The goal is to determine the metrics to study the impacts of long delays on TCP performance.

cs.NI↗

arXiv2013

FavourQueue: a Parameterless Active Queue Management to Speed Up Short TCP Flows (and others too!)

This paper presents and analyses the implementation of a novel active queue management (AQM) named FavorQueue that aims to improve delay transfer of short lived TCP flows over best-effort networks. The idea is to dequeue packets that do not belong to a flow previously enqueued first. The rationale is to mitigate the delay induced by long-lived TCP flows over the pace of short TCP data requests and to prevent dropped packets at the beginning of a connection and during recovery period. Although the main target of this AQM is to accelerate short TCP traffic, we show that FavorQueue does not only improve the performance of short TCP traffic but also improves the performance of all TCP traffic in terms of drop ratio and latency whatever the flow size. In particular, we demonstrate that FavorQueue reduces the loss of a retransmitted packet, decreases the number of dropped packets recovered by RTO and improves the latency up to 30% compared to DropTail. Finally, we show that this scheme remains compliant with recent TCP updates such as the increase of the initial slow-start value.

cs.NI↗

arXiv2007

TCP throughput guarantee in the DiffServ Assured Forwarding service: what about the results?

Since the proposition of Quality of Service architectures by the IETF, the interaction between TCP and the QoS services has been intensively studied. This paper proposes to look forward to the results obtained in terms of TCP throughput guarantee in the DiffServ Assured Forwarding (DiffServ/AF) service and to present an overview of the different proposals to solve the problem. It has been demonstrated that the standardized IETF DiffServ conditioners such as the token bucket color marker and the time sliding window color maker were not good TCP traffic descriptors. Starting with this point, several propositions have been made and most of them presents new marking schemes in order to replace or improve the traditional token bucket color marker. The main problem is that TCP congestion control is not designed to work with the AF service. Indeed, both mechanisms are antagonists. TCP has the property to share in a fair manner the bottleneck bandwidth between flows while DiffServ network provides a level of service controllable and predictable. In this paper, we build a classification of all the propositions made during these last years and compare them. As a result, we will see that these conditioning schemes can be separated in three sets of action level and that the conditioning at the network edge level is the most accepted one. We conclude that the problem is still unsolved and that TCP, conditioned or not conditioned, remains inappropriate to the DiffServ/AF service.

cs.NI↗