Introduction Jacobs University, Bremen Leone Project: leone-project.eu Supported by: Oct 11, 2016 University of Ghent, Belgium Prof. Dr. Filip De Turck Jacobs University Bremen, Germany Dr. Kinga Lipskoch Jacobs University Bremen, Germany Prof. Dr. Jürgen Schönwälder Dissertation Committee TUM Seminar, Raitenhaslach Vaibhav Bajpai Motivation using Large Scale Measurement Platforms Understanding the Impact of Network Infrastructure Changes Q/A Takeaway Results Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 1 / 26 Flamingo Project: fmamingo-project.eu
Introduction Methodology Tiis thesis would not have been possible without these amazing people! Motivation Takeaway Results Q/A Introduction Web Similarity Web Similarity Part II Part I Research Contributions 2 / 26 • • – – – – – • What’s ¡missing: ¡Many ¡things, ¡but ¡in
Introduction Takeaway SamKnows [4], BISmark [5], RIPE Atlas [6, 4], PerfSONAR [4] et al. CAIDA Ark [1], DIMES [2], iPlane [3] et al. measurement tests to satisfy specifjc use-case requirements. An infrastructure of dedicated hardware probes that periodically runs Internet Motivation Q/A Introduction | Motivation Results Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 3 / 26 ▶ Large-scale Internet Measurement Platform ▶ Use-cases Since ′ 98: Topology Mapping (Mature) Since ′ 08: Network Performance ▶ Measure performance and reliability of broadband access networks ▶ Facilitate regulators to make better policy decisions.
Introduction Results 2. Measuring Access Network Performance 1. Measuring IPv6 Performance Introduction | Motivation Motivation Takeaway Q/A Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 4 / 26 ▶ Tiis dissertation expands the goal: Since ′ 98: Topology Mapping Since ′ 08: Network Performance ▶ Measure performance and reliability of broadband access networks ▶ Facilitate regulators to make better policy decisions. ▶ Understand the impact of network infrastructure changes
Introduction Takeaway * entries are papers currently under review. 3. Measuring Access Network Performance Motivation 2. Measuring IPv6 Performance Introduction | Research Contributions Q/A 5 / 26 Results Methodology Introduction Web Similarity Web Similarity Part II Research Contributions Part I [ COMST ′ 15 ] 1. Survey on Internet Performance Measurement Platforms ▶ Measuring TCP Connect Times [ NETWORKING ′ 15 ] ▶ Measuring YouTube Performance [ PAM ′ 15 ] ▶ Measuring Efgects of Happy Eyeballs [ ANRW ′ 16 ] [ CNSM ′ 16 ] ▶ Measuring Web Similarity ▶ RIPE Atlas Vantage Point Selection [ ∗ ] ▶ Dissecting Last-mile Latency Characteristics [ ∗ ] ▶ Lessons Learned from using RIPE Atlas [ CCR ′ 15 ]
Introduction Dasu IETF xrblock BBF IEEE ITU-T SamKnows BISmark Netradar IETF LMAP Portolan RIPE Atlas perfSONAR Benoit Donnet et al. [7] Hamed Haddadi et al. [8] Benoit Donnet et al. [9] Bajpai et al. [4] IETF IPPM Topology Discovery Motivation Results Research Contributions Part I Part II Web Similarity Web Similarity Introduction Methodology Takeaway Mobile Access Q/A Part I Performance Measurements Fixed-line Access Operational Support Standardization Efgorts Internet Measurement Platforms 6 / 26
Introduction Part II | Overview Bajpai et al. [13] Eravuchira et al. [12] Bajpai et al. [11] Ahsan et al. [10] Measuring IPv6 Performance Web Similarity YouTube Happy Eyeballs TCP Connect Times Q/A Motivation Takeaway Results Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 7 / 26
Introduction Takeaway ARIN LACNIC RIPE APNIC Motivation Part II | Motivation Q/A 8 / 26 Results Methodology Introduction Web Similarity Research Contributions Part I Part II Web Similarity ▶ Literature has largely focussed on measuring IPv6 adoption [14, 15, 16] ( ′ 10 − ′ 14). ▶ Addressing ▶ Naming ▶ Routing ▶ Reachability ▶ Very little work [17] on measuring performance of service delivery over IPv6. ▶ Largely due to lack of available content over IPv6. ▶ A number of signifjcant events occured during the span of this dissertation. Apr ′ 11 ▶ IANA IPv4 Address Exhaustion [18] ▶ World IPv6 Day ′ 11 [19] Sep ′ 12 ▶ World IPv6 Launch Day ′ 12 [20] Jun ′ 14 ▶ RIR IPv4 Address Exhaustion [18] Sep ′ 15
Introduction Motivation 1 Comcast, Deutsche Telekom AG, AT&T, Verizon Wireless, T-Mobile USA 21.41% Germany 23.62% United States 27.38% Switzerland 40.49% Belgium 05/2016 11.48% 09/2012 0.85% 9 / 26 Part II | Motivation Q/A Takeaway Results Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions ▶ Large IPv6 broadband rollouts 1 [13]. ▶ Global IPv6 adoption [21]. ▶ Tiis study closes the gap. ▶ It measures IPv6 performance of operational dual-stacked content delivery services.
Introduction TCP Connect Times * entries are papers currently under review. Bajpai et al. [13] Eravuchira et al. [12] Bajpai et al. [11] Ahsan et al. [10] Measuring IPv6 Performance Web Similarity YouTube Happy Eyeballs Part II | Measuring Web Similarity Motivation Q/A Takeaway Results Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 10 / 26
Introduction Web Similarity | Introduction We want to know: Motivation similarity over IPv4 / IPv6. No study comparing web over IPv4 and IPv6. of dual-stacked websites has compared performance Recent work [13], [17], [15] 11 / 26 Q/A Web Similarity Takeaway Research Contributions Part II Part I Web Similarity Introduction Methodology Results Websites with AAAA entries 10.0% 8.0% ALEXA 1M Websites W6D W6LD 6.0% 4.0% 2.0% 0.0% 2010 2011 2012 2013 2014 2015 2016 http://www.employees.org/ ∼ dwing/aaaa-stats ▶ How similar are webpages accessed over IPv6 to their IPv4 counterparts? ▶ What factors contribute to the dissimilarity over IPv4 and IPv6?
Introduction Takeaway To the best of our knowledge, this is the fjrst study to: 4. Both same-origin and cross-origin sources contribute to the failure rates over IPv6. 3. Failure rates over IPv6 are largely due to DNS resolution error on images, js and CSS. 2. Websites (27%) have some fraction of webpage elements failing over IPv6. We measure against ALEXA top 100 dual-stacked websites. Web Similarity | Introduction Motivation Q/A Results Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 12 / 26 1. simweb : A tool for measuring web similarity over IPv4 and IPv6. ▶ Measure webpage similarity over IPv4 and IPv6. ▶ Investigate IPv6 adoption that goes beyond the root page of a dual-stacked website.
Introduction Takeaway Tie number of same-origin & cross-origin sources. 2. Service Complexity Tie number & size of fetched webpage elements. 1. Content Complexity We use 2 well-known webpage complexity metrics from literature [22, 23]: Web Similarity | Methodology Q/A Results Motivation Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 13 / 26
Introduction Methodology as measurement targets [13]. Motivation Q/A Takeaway Results Web Similarity | Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions 14 / 26 1. www.google.com 2. www.facebook.com 3. www.youtube.com 4. www.yahoo.com ▶ We use the ALEXA top 100 dual-stacked websites 5. www.wikipedia.org 6. www.qq.com 7. www.blogspot.com 8. …
Introduction Methodology Motivation Web Similarity | Methodology Q/A Takeaway Results 15 / 26 Introduction Web Similarity Web Similarity Part II Part I Research Contributions ALEXA Dual-Stacked Tie simweb test: Top 100 ▶ runs twice (once for each AF). HTTP GET IPv6 simweb IPv4 results ▶ repeats every hour. DSL/Cable SamKnows Modem Tests ▶ uses user-agent string: Mozilla/4.0 Probe HTTPS POST Data Collector
Introduction Results We measure from 80 dual-stacked SamKnows probes. Motivation Web Similarity | Methodology Q/A Takeaway 16 / 26 Methodology Introduction Web Similarity Web Similarity Part II Part I Research Contributions NETWORK TYPE # RESIDENTIAL 55 NREN / RESEARCH 11 BUSINESS / DATACENTER 09 OPERATOR LAB 04 IXP 01 RIR # RIPE 42 ARIN 29 APNIC 07 AFRINIC 01 LACNIC 01
Recommend
More recommend
