Workshop on Application Layer Traffic Optimization (ALTO)

NEC Laboratories Europe, Heidelberg, 24 November 2010



  1. Tobias Hoßfeld: "Future Internet Applications & Overlays"
    A short introduction of the FIA research group is given.
    Slides


  2. Saverio Niccolini: "RTC group intro and overview of activities in the P2P area"
    A brief overview of NAPA-WINE and activities in COAST, CDN is given.


  3. Jan Seedorf: "ALTO Simulations"
    Application Layer Traffic Optimization (ALTO) is a means for operators to guide the resource provider selection of distributed applications. We study the potential cost-savings for operators through ALTO-guidance for a specific type of P2P application: P2P Live Streaming. We use datasets that model the Internet's AS-level routing topology with high accuracy and which estimate the business relationships between connected ASes on the Internet. Based on this data, we investigate different ALTO strategies and quantify the number of costly AS-hops traversed. Our results show that indeed transmission costs can be reduced significantly for P2P Live Streaming with ALTO. However, for this particularly delay-sensitive type of application, ISPs have to be careful not to over-localize traffic: if peers connect to too many peers which are in the same AS but have low upload capacity, chunk loss increases considerably(resulting in poor video quality).


  4. Frank Lehrieder: "Characterizations of BitTorrent Swarms and their Distribution in the Internet"
    BitTorrent is used to distribute a large number of different files in today's Internet. However, the research community lacks appropriate characterizations of those BitTorrent swarms, e.g. of the number of users exchanging the same file and how they are distributed across different autonomous systems. This talk presents the results of our measurement study which provides insights into the nature of real-life BitTorrent swarms.
    Slides


  5. Frank Lehrieder: "The Impact of Caching on BitTorrent-Like P2P Systems"
    Despite legal issues, caching of BitTorrent traffic is applied by a large set of Internet service providers in order to reduce inter-domain traffic. The aim of this talk is to present a mathematical model which captures the impact of caching on BitTorrent swarms and the corresponding inter-domain traffic. From the model, we'll derive some practical implications for the use of caches.
    Slides


  6. Simon Oechsner: "Locality-awareness in BitTorrent"
    Locality-awareness is considered as a promising approach to increase the efficiency of content distribution by peer-to-peer (P2P) networks, e.g., BitTorrent. It is intended to reduce the inter-domain traffic which is costly for Internet service providers (ISPs) and simultaneously increase the performance from the viewpoint of the P2P users, i.e, shorten download times. In this talk, we discuss results studying locality-aware-implementations considering real-life, skewed peer distributions and heterogeneous access bandwidths of peers.
    Slides


  7. Armin Jahanpanah, Jan Seedorf: "ALTO Client and ALTO Server Implementation with demo"
    Following ongoing IETF standardisation, we have implemented a prototypical ALTO client and ALTO server. The ALTO server is a stand-alone application which hosts the logic to guide applications to optimize network resource usage; the ALTO client provides an API for interacting with the ALTO server. In this talk, we give and overview of our software implementation and the corresponding message flow. In addition, we show a live ALTO demo where an ALTO-ranking is provided for arbitrary IP-ranges.


  8. Mischa Schmidt, Stefano Napolitano, Andrea Cavaliere: "Large-scale Bittorrent Trials: Testbed, concept, and first results"
    To validate the ALTO concept in a large-scale trial with real implementations, we are conducting several trials. Specifically, the experiments conducted in the trials are aimed at answering the following questions: a) To what extent can an operator reduce intra-area or backbone transmission costs with ALTO while still fulfilling application requirements (e.g. low chunk loss and low delay for P2P live streaming)?, b) To what extent can the ALTO concept be used to improve the applications Quality-of-Service, i.e., to what extent can ALTO reduce application layer delay or chunk loss?, c) What is the performance and scalability of the ALTO components developed within Napa-Wine, i.e., are the implementations able to work efficiently in a real-world scenario? Initial results validate the ALTO concept developed within NAPA-WINE for BitTorrent filesharing. These results demonstrate for different scenarios to what extent backbone traffic transmission costs can be reduced with ALTO.


  9. Armin Jahanpanah, Jan Seedorf: "P2P Live Streaming ALTO Trials: Concept and status quo"
    To test the Napa-Wine p2p live streaming client software in conjunction with ALTO functionality, the following scenario is envisioned: The Napa-Wine software uses the integrated ALTO client to connect to an ALTO server which is operated by an ISP. In the case of Napa-Wine, the operator hosting the ALTO server is Polish Telecom (TP). In order to deploy an ALTO server in the TP network, TP provided provisioning data from its network into the ALTO serverthrougha dedicated ALTO provisioning interface. This interface converts data sets from the operator such as operational policies, costs among links, or similar information into a format which can be imported into the ALTO-Server. This talk provides an overview of the ongoing ALTO live streaming trials and the status quo of technical issues.


  10. Thomas Zinner: "Towards a QoE-aware P2P VoD"
    The scalable extension of H.264/AVC offers the possibility to adopt the user perceived video quality to technical parameters such as application- and network level Quality of Service. A proper quality control mechanism needs to take the main influence parameters into account. This requires an understanding of the impact of uncontrollable influence factors such as packet loss and controllable influence factors such as a lower frame rate or a lower resolution. In the first part of this talk we discuss the results of our investigation on video quality for the scalable extension of H.264/AVC. After that we discuss the implementation of a P2P based VOD system using this codec and show results of our performance evaluation.
    Slides