Att en hemsida tar extra lång tid att ladda är inte ovanligt och kan bero på att trafiken från en klient tar en helt annan väg till hemsidans server än den som är geografiskt närmast. Orsaken bakom problemet är att DNS-förfrågningarna färdas onödigt långa sträckor. NetNod är ett företag som tillhandahåller dessa internettjänster, bland annat rotservrar runt om i världen. De vill ta reda på varför trafik från olika klienter inte alltid går den geografiskt närmaste vägen till deras anycastnoder Problemställningen för examensarbetet är att analysera varifrån trafik till företagets anycastnod i Miami kommer. Realiseringen av detta gjordes genom att utveckla en programkod som markerade geografisk placering av olika klienter utifrån deras IP-adress. Förinspelade trafikdata från NetNod analyserades. Detta gjordes för att visa på vilka problem som observeras i peering och anycastrouting mellan internetprotokoll. Resultatet redovisades med en karta med markeringar av de IP-adresser där deras trafik analyserades för att se hur det transporterades till anycastnoden. Utifrån detta har resultatet visat på vilka avvikelser och mönster som uppstått inom BGP-routing när trafiken färdas till anycastnoden. De avvikelser som hittats är hur olika routingregler manipulerat trafikens transport till anycastnoden och gör att trafiken från klienterna inte tar den geografiskt närmaste vägen till anycastnoden. / The fact that a website takes an unusually long time to load is not uncommon. This can be due to a client taking a different path to the websites server than one that is geographically closer. One reason behind this problem is that DNS-queries travel unnecessarily long distances. NetNod is a company that provides internet services and maintains one of the few root-servers around the world. The company wants to know why traffic from different clients do not always go via the geographically closest route to anycast nodes. The objective of the thesis is to analyze where traffic to NetNods anycast node in Miami geographically originates from. In order to do this, a computer program was developed in which plots the geographical location of different clients from their IP-address. Pre-recorded data from the company was used as a data source for the program. This was done to show different challenges in peering and anycast routing between internet protocols. The result is presented via a map with plots of where the IP-addresses are geographically coming from to the anycast node in Miami, it was generated by the developed program. The generated map showed anomalies and patterns of how the traffic is transported in large junctions as well as how routing rules are applied, this is one reason to why the traffic does not always go the geographically closest route.

Cloud computing platforms allow users to deploy geographically distributed applications on servers around the world. Applications may be simple to deploy on these platforms, but it is up to the user and the application to decide which regions and servers to use for application placement. Furthermore, network conditions and routing between the geo-distributed servers change over time, which can lead to sub-optimal performance of applications deployed on such servers. A user could either employ a static deployment configuration of servers, or attempt to use a more dynamic configuration. However, both have inherent limitations. A static configuration will be sub-optimal, as it will be unable to adapt to changing network conditions. A more dynamic approach where an application could switch over or transition to a more suitable server could be beneficial, but this can be very complex in practice. Furthermore, such a solution is more about adapting to change as it happens, and not beforehand. This thesis will investigate the possibility of forecasting impending routing changes between servers, by leveraging messages generated by the Border Gateway Protocol (BGP) and past knowledge about routing changes. BGP routers can delay BGP updates due to factors such as the minimum route advertisement interval (MRAI). Thus, out proposed solution involves forwarding BGP updates downstream in the network, before BGP routers process them. As routing between servers changes, so does the latency, meaning that the latency then could be predicted to some degree. This observation could be applied to realize when the latency to a server increases or decreases past another server. This in turn facilitates the decision process of selecting the most optimal servers in terms of latency for application deployment. The solution presented in this thesis can successfully predict routing changes between end-points in an enclosed environment, and inform users ahead of time that the latency is about to change. The time gained by such predictions depend on factors such as the number of ASs between the end-points, the MRAI, and the update processing delay imposed on BGP routers. Time gains between tens of milliseconds to over 2 minutes has been observed.

The delivery of IP traffic on the Internet depends on the complex interactions between a set of autonomous systems that exchange routing information about IP prefix destinations utilizing the Border Gateway Protocol (BGP). Autonomous systems are often connected to a route server located at an Internet eXchange Point, which facilitates the administration of BGP peering arrangements for all parties connected to it. One of the most popular open-source implementations of BGP is the FRRouting software suite, making it an important part of the Internet infrastructure. This thesis investigates the performance of FRRouting, configured as a route server, in terms of its capabilities of announcing routing information on the network to a set of peers emulating autonomous systems. The routing information consists of a set of distinct IP prefixes that FRRouting receives from its peers. With various benchmarks of different configurations, we relate the number of received prefixes to the number of prefixes that FRRouting has announced on the network to its peers in a given time span. The output of the thesis is a wide overview of how the performance of FRRouting is impacted by different configurations such as filtering of specific prefixes that are not announced to the peering networks. The obtained results show that there exists a divergence between the number of prefixes that have been received and announced by FRRouting. Specifically, the discrepancy shows that FRRouting, in our benchmarks, is incapable of announcing prefixes at the same rate as it receives these prefixes from its peers. In general, the number of announced prefixes is dependent on how the prefix filter is configured. However, one can question what real-world limitations this may result in. Suggestions for future work include developing a more realistic benchmarking environment that does not rely on emulated peers as well as improving how the routing information is recorded. There also exists a wide variety of other metrics and configurations of FRRouting that may reveal further limitations. / Leveransen av IP-trafik på Internet beror av komplexa interaktioner mellan en uppsättning autonoma system som utbyter dirigeringsinformation med hjälp av Border Gateway Protocol (BGP). Autonoma system är ofta anslutna till en dirigeringsserver belägen vid en Internetknutpunkt vilket underlättar administrationen av BGP-förbindelser mellan de parter som är anslutna till dirigeringsservern. En av de mest populära implementationerna av BGP med öppen källkod är FRRouting vilket gör denna mjukvara till en betydelsefull komponent för Internets infrastruktur. Detta arbete undersöker prestandan av FRRouting konfigurerad som en dirigeringsserver vad gäller dess kapacitet att behandla och via nätverket vidarebefordra dirigeringsinformation till en uppsättning autonoma system. Dirigeringsinformationen består av en samling IP-prefix som FRRouting erhåller från de autonoma systemen. Genom att variera konfigurationen av FRRouting undersöker vi hur antalet mottagna IP-prefix relaterar till den mängd IP-prefix som FRRouting har vidarebefordrat till de autonoma systemen under en given tidsperiod. Arbetet resulterar i en bred genomgång av hur prestandan för FRRouting påverkas av olika konfigurationer såsom filtrering av specifika prefix. De erhållna resultaten visar att antalet vidarebefordrade IP-prefix skiljer sig markant från antalet prefix som dirigeringsservern erhållit från de autonoma systemen. Denna avvikelse visar att FRRouting inte är kapabel att vidarebefordra IP-prefix i samma takt som dessa mottages från de autonoma systemen. I allmänhet beror antalet vidarebefordrade IP-prefix av hur prefixfiltreringen konfigurerats. Det kan dock ifrågasättas vilka verkliga begränsningar detta kan resultera i. Förslag på framtida arbeten inkluderar att utveckla en förbättrad testmiljö som inte förlitar sig på emulerade autonoma system samt att förbättra tekniken för insamling av vidarebefordrade IP-prefix. Det existerar även ett stort antal andra mätvärden och konfigurationer av FRRouting som möjligtvis kan resultera i att ytterligare begränsningar uppdagas.

This dissertation presents measures of efficiency and locality for Internet addressing schemes. Historically speaking, many issues, faced by the Internet, have been solved just in time, to make the Internet just work~\cite{justWork}. Consensus, however, has been reached that today's Internet routing and addressing system is facing serious scaling problems: multi-homing which causes finer granularity of routing policies and finer control to realize various traffic engineering requirements, an increased demand for provider-independent prefix allocations which injects unaggregatable prefixes into the Default Free Zone (DFZ) routing table, and ever-increasing Internet user population and mobile edge devices. As a result, the DFZ routing table is again growing at an exponential rate. Hierarchical, topology-based addressing has long been considered crucial to routing and forwarding scalability. Recently, however, a number of research efforts are considering alternatives to this traditional approach. With the goal of informing such research, we investigated the efficiency of address assignment in the existing (IPv4) Internet. In particular, we ask the question: ``how can we measure the locality of an address scheme given an input AS-level graph?'' To do so, we first define a notion of efficiency or locality based on the average number of bit-hops required to advertize all prefixes in the Internet. In order to quantify how far from ``optimal" the current Internet is, we assign prefixes to ASes ``from scratch" in a manner that preserves observed semantics, using three increasingly strict definitions of equivalence. Next we propose another metric that in some sense quantifies the ``efficiency" of the labeling and is independent of forwarding/routing mechanisms. We validate the effectiveness of the metric by applying it to a series of address schemes with increasing randomness given an input AS-level graph. After that we apply the metric to the current Internet address scheme across years and compare the results with those of compact routing schemes.

<p>Arbetet tar upp BGP-multihoming för mindre organisationer. Den jämför ett kommersiell alternativ mot open source-alternativ. Detta är för att se vad som krävs en av open source-lösning för ge konkurrens på routermarknaden, samt besparingar för organisationer som ska använda denna lösning.Praktiska och teoretiska jämförelser görs där Cisco-lösningar jämförs med OpenBGPD. Datan utifrån dessa tester används för att svara på problemfrågan. Sammanfattningsvis har open source produkter en fördel hårdvarumässigt på grund av lägre kostnader medan kommersiella har stora fördelar då de förlitar sig på nyare standarder.</p>

This document specifies a simple signalling mechanism for inter-domain QoS marking using a BGP Extended Community QoS Attribute. Class based packet forwarding for delay and loss critical services is currently performed in an individual AS internal manner. The new QoS marking attribute makes the QoS class setup within the IP prefix advertising AS known to all access and transit ASes. This enables individual (re-)marking and forwarding treatment adaptation to the original QoS class setup of the respective IP prefix. The attribute provides the means to signal QoS markings on different layers, which are linked together in QoS class sets. It provides inter-domain and cross-layer insight into the QoS class mapping of the source AS with minimal signalling traffic.

The increasing number of delay and loss critical services in packet networks require differentiated packet handling in the forwarding plane. Quality of Service (QoS) guarantees can be given for networks using resource reservation and admission control. However, such strategies require complex control plane extensions and might lead to higher operation expenditures. Network operators therefore often use over-provisioning and traffic differentiation to offer cheaper class of service quality in their internet protocol (IP) packet networks. Priority marking and forwarding of packetized data traffic can be realized mainly using different layer two and three mechanisms. The number of differentiated classes and their autonomous system (AS) internal implementation is at the operator’s choice. This paper proposes a concept of cross-domain and cross-layer priority signaling between packet switched networks to be used at the inter-AS peering points. It makes use of the Border Gateway Protocol (BGP) as the predominantly used routing protocol for AS peering communication. A new BGP-4 path attribute is used to convey the structured priority information. The new concept ensures that all receiving AS peers are consistently and comprehensively informed about the QoS handling within the IP prefix originating AS. Based on this information, all ASes can perform close QoS treatment approximation in a cross-domain and cross-layer manner. The approach is now work in progress at the IETF.

This document focuses on Class of Service Interconnection at inter- domain peering points. It specifies two new non-transitive attributes, which enable adjacent peers to signal Class of Service Capabilities and certain Class of Service admission control Parameters. The new "CoS Capability Attribute" is deliberately kept simple and denotes the general EF, AF Group and BE forwarding support across the advertising AS. The second "CoS Parameter Attribute" is of variable length and contains a more detailed description of available forwarding behaviours using the PHB id Code encoding. Each PHB id Code is associated with rate and size based traffic parameters, which will be applied in the ingress AS Border Router for admission control purposes to a given forwarding behaviour. The denoted Class of Service forwarding support is meant as the AS externally available (transit) Class of Service support.

Mit der zunehmenden Popularität des Internets steigt die Anzahl der Nutzer und vor allem die Anzahl zeit- und verlustkritische Dienste – wie zum Beispiel „Voice over IP“, Videoübertragungen und netzbasierte Spiele. Das Internet ist dabei der Zusammenschluss von ca. 30.000 Betreibernetzen, die mit Hilfe des „Internet Protocol (IP)“ derzeit ohne jede Dienstgüteunterstützung den Datenverkehraustausch realisieren. Massive Überdimensionierung der Netzkapazitäten führen zu einer Netzauslastung von nur ca. 10% und entsprechend guter Übertragungsqualität. Mit steigendem Verkehrsaufkommen wird in dieser Dissertation erwartet, das die Netzbetreiber infolge des Kostendrucks nicht schritthaltend den überhöhten Netzausbau aufrechterhalten können und somit Qualitätseinbußen zu erwarten sind. Innerhalb der Betreiber wird bereits jetzt Verkehrstrennung betrieben, jedoch am Übergabepunkt verworfen und im besten Fall im Nachbarnetz durch aufwendige Analyse erneut vorgenommen. Im Rahmen dieser Arbeit wurde deshalb ein domänen- und schichtenübergreifendes Konzept zur Realisierung grob-granularer Dienstgüte in IP-Netzen entworfen, zur Standardisierung bei der „Internet Engineering Task Force (IETF)“ vorgeschlagen, implementiert und in Auszügen simuliert und getestet. Dabei werden die Verkehrsklasseninformationen mehrere Netzschichten in transitiven Nachrichtenelementen des „Border Gateway Protocol (BGP)“ signalisiert und schichtenübergreifend assoziiert. Die vorliegende Dissertation beinhaltet im wesentlichen drei Teile: 1. Eine umfassende Zusammenstellung von vorhandenen Dienstgütekonzepten einschließlich der bereits existierenden QoS-Funktionselemente in verfügbaren Netzelementen, 2. Die detaillierte Spezifikation des neuen Konzeptes und 3. den Ergebnissen der Simulations- und Implementierungsaktivitäten zum Nachweis der Funktion und Skalierbarkeit des Entwurfes. Zwei wesentliche Erkenntnisse und Forderungen sind durch die Bearbeitung des Themas erwachsen. Die Einfachheit der Konzeptstruktur und die Einfachheit der angestrebten Dienstgüteunterstützung. Die angestrebte Dienstgüte beschränkt sich deshalb auf die primitive Verkehrstrennung in mehrere Klassen, die in den Weiterleitungsknoten getrennt abgelegt und mit verschiedenem Vorrang behandelt werden.

Arbetet tar upp BGP-multihoming för mindre organisationer. Den jämför ett kommersiell alternativ mot open source-alternativ. Detta är för att se vad som krävs en av open source-lösning för ge konkurrens på routermarknaden, samt besparingar för organisationer som ska använda denna lösning.Praktiska och teoretiska jämförelser görs där Cisco-lösningar jämförs med OpenBGPD. Datan utifrån dessa tester används för att svara på problemfrågan. Sammanfattningsvis har open source produkter en fördel hårdvarumässigt på grund av lägre kostnader medan kommersiella har stora fördelar då de förlitar sig på nyare standarder.