Anycast Latency: How Many Sites Are Enough?

  • Ricardo de Oliveira SchmidtEmail author
  • John Heidemann
  • Jan Harm Kuipers
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10176)


Anycast is widely used today to provide important services such as DNS and Content Delivery Networks (CDNs). An anycast service uses multiple sites to provide high availability, capacity and redundancy. BGP routing associates users to sites, defining the catchment that each site serves. Although prior work has studied how users associate with anycast services informally, in this paper we examine the key question how many anycast sites are needed to provide good latency, and the worst case latencies that specific deployments see. To answer this question, we first define the optimal performance that is possible, then explore how routing, specific anycast policies, and site location affect performance. We develop a new method capable of determining optimal performance and use it to study four real-world anycast services operated by different organizations: C-, F-, K-, and L-Root, each part of the Root DNS service. We measure their performance from more than 7,900 vantage points (VPs) worldwide using RIPE Atlas. (Given the VPs uneven geographic distribution, we evaluate and control for potential bias.) Our key results show that a few sites can provide performance nearly as good as many, and that geographic location and good connectivity have a far stronger effect on latency than having many sites. We show how often users see the closest anycast site, and how strongly routing policy affects site selection.


Median Latency High Latency Open Resolver Content Delivery Network Good Latency 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.



We thank Geoff Huston (APNIC), George Michaelson (APNIC), Ray Bellis (ISC),Cristian Hesselman (SIDN Labs), Benno Overeinder (NLnet Labs) and Jaap Akkerhuis (NLnet Labs), Duane Wessels (Verisign), Paul Vixie (Farsight), Romeo Zwart (RIPE NCC), Anand Buddhdev (RIPE NCC), and operators from C Root for their technical feedback.

This research uses measurements from RIPE Atlas, operated by RIPE NCC.

Ricardo Schmidt’s work is in the context of SAND (Self-managing Anycast Networks for the DNS: and DAS (DNS Anycast Security: projects, sponsored by SIDN, NLnet Labs and SURFnet.

John Heidemann’s work is partially sponsored by the U.S. Dept. of Homeland Security (DHS) Science and Technology Directorate, HSARPA, Cyber Security Division, via SPAWAR Systems Center Pacific under Contract No. N66001-13-C-3001, and via BAA 11-01-RIKA and Air Force Research Laboratory, Information Directorate under agreement numbers FA8750-12-2-0344 and FA8750-15-2-0224. The U.S. Government is authorized to make reprints for Governmental purposes notwithstanding any copyright. The views contained herein are those of the authors and do not necessarily represent those of DHS or the U.S. Government.


  1. 1.
    Abley, J., Lindqvist, K.E.: Operation of Anycast Services. RFC 4786 (2006)Google Scholar
  2. 2.
    Akhtar, Z., Hussain, A., Katz-Bassett, E., Govindan, R.: DBit: assessing statistically significant differences in CDN performance. In: IFIP TMA (2016)Google Scholar
  3. 3.
    Bajpai, V., Eravuchira, S.J., Schönwälder, J.: Lessons learned from using the RIPE atlas platform for measurement research. ACM CCR 45(3), 35–42 (2015)CrossRefGoogle Scholar
  4. 4.
    Ballani, H., Francis, P.: Towards a global IP anycast service. In: ACM SIGCOMM, pp. 301–312 (2005)Google Scholar
  5. 5.
    Ballani, H., Francis, P., Ratnasamy, S.: A measurement-based deployment proposal for IP anycast. In: ACM IMC, pp. 231–244 (2006)Google Scholar
  6. 6.
    Bellis, R.: Researching F-root Anycast Placement Using RIPE Atlas (2015).
  7. 7.
    Boothe, P., Bush, R.: Anycast Measurements Used to Highlight Routing Instabilities. NANOG 34 (2005)Google Scholar
  8. 8.
    Brownlee, N., Claffy, K.C., Nemeth, E.: DNS Root/gTLD performance measurement. In: USENIX LISA, pp. 241–255 (2001)Google Scholar
  9. 9.
    Brownlee, N., Ziedins, I.: Response time distributions for global name servers. In: PAM (2002)Google Scholar
  10. 10.
    Bush, R.: DNS anycast stability: some initial results. In: CAIDA/WIDE Workshop (2005)Google Scholar
  11. 11.
  12. 12.
    Calder, M., Fan, X., Hu, Z., Katz-Bassett, E., Heidemann, J., Govindan, R.: Mapping the expansion of Google’s serving infrastructure. In: ACM IMC, pp. 313–326 (2013)Google Scholar
  13. 13.
    Calder, M., Flavel, A., Katz-Bassett, E., Mahajan, R., Padhye, J.: Analyzing the performance of an anycast CDN. In: ACM IMC, pp. 531–537 (2015)Google Scholar
  14. 14.
    Castro, S., Wessels, D., Fomenkov, M., Claffy, K.: A day at the root of the internet. ACM CCR 38(5), 41–46 (2008)CrossRefGoogle Scholar
  15. 15.
    Cicalese, D., Augé, J., Joumblatt, D., Friedman, T., Rossi, D.: Characterizing IPv4 anycast adoption and deployment. In: ACM CoNEXT (2015)Google Scholar
  16. 16.
    Cicalese, D., Joumblatt, D., Rossi, D., Buob, M.-O., Augé, J., Friedman, T.: A fistful of pings: accurate and lightweight anycast enummeration and geolocation. In: IEEE INFOCOM, pp. 2776–2784 (2015)Google Scholar
  17. 17.
    Colitti, L.: Effect of anycast on K-root. In: 1st DNS-OARC Workshop (2005)Google Scholar
  18. 18.
    DNS Root Servers.
  19. 19.
    Fan, X., Heidemann, J., Govindan, R.: Evaluating anycast in the domain name system. In: IEEE INFOCOM, pp. 1681–1689 (2013)Google Scholar
  20. 20.
    Fan, X., Katz-Bassett, E., Heidemann, J.: Assessing affinity between users and CDN sites. In: Steiner, M., Barlet-Ros, P., Bonaventure, O. (eds.) TMA 2015. LNCS, vol. 9053, pp. 95–110. Springer, Heidelberg (2015). doi: 10.1007/978-3-319-17172-2_7 Google Scholar
  21. 21.
    Fomenkov, M., Claffy, K.C., Huffaker, B., Moore, D.: Macroscopic internet topology and performance measurements from the DNS root name servers. In: USENIX LISA, pp. 231–240 (2001)Google Scholar
  22. 22.
  23. 23.
    Kuipers, J.H.: Analysing the K-root anycast infrastructure (2015).
  24. 24.
    Lee, B.-S., Tan, Y.S., Sekiya, Y., Narishige, A., Date, S.: Availability and effectiveness of root DNS servers: a long term study. In: IFIP/IEEE NOMS, pp. 862–865 (2010)Google Scholar
  25. 25.
    Lee, T., Huffaker, B., Fomenkov, M., Claffy, K.C.: On the problem of optimization of DNS root servers’ placement. In: PAM (2003)Google Scholar
  26. 26.
    Liang, J., Jiang, J., Duan, H., Li, K., Wu, J.: Measuring query latency of top level DNS servers. In: Roughan, M., Chang, R. (eds.) PAM 2013. LNCS, vol. 7799, pp. 145–154. Springer, Heidelberg (2013). doi: 10.1007/978-3-642-36516-4_15 CrossRefGoogle Scholar
  27. 27.
    Liu, Z., Huffaker, B., Fomenkov, M., Brownlee, N., Claffy, K.C.: Two days in the life of the DNS anycast root servers. In: Uhlig, S., Papagiannaki, K., Bonaventure, O. (eds.) PAM 2007. LNCS, vol. 4427, pp. 125–134. Springer, Heidelberg (2007). doi: 10.1007/978-3-540-71617-4_13 CrossRefGoogle Scholar
  28. 28.
    Palsson, B., Kumar, P., Jafferalli, S., Kahn, Z.A.: TCP over IP anycast - pipe dream or reality? (2015).
  29. 29.
    Pang, J., Hendricks, J., Akella, A., Prisco, R.D., Maggs, B., Seshan, S.: Availability, usage, and deployment characteristics of the domain name server. In: ACM IMC, pp. 1–14 (2004)Google Scholar
  30. 30.
    Partridge, C., Mendez, T., Milliken, W.: Host Anycasting Service. RFC 1546 (1993)Google Scholar
  31. 31.
    RIPE NCC. Dnsmon (2015).
  32. 32.
    RIPE NCC Staff: RIPE Atlas: a global Internet measurement network. Internet Protocol J. 18(3), 2–26 (2015)Google Scholar
  33. 33.
    Rootops. Events of 2015–11-30. Technical report, Root Server Operators (2015)Google Scholar
  34. 34.
    Sarat, S., Pappas, V., Terzis, A.: On the use of anycast in DNS. In: ICCCN, pp. 71–78 (2006)Google Scholar
  35. 35.
    Schmidt, R.d.O., Heidemann, J., Kuipers, J.H.: Anycast latency: how many sites are enough? Technical report ISI-TR-2016-708, USC-ISI, May 2016Google Scholar
  36. 36.
    Spring, N., Mahajan, R., Anderson, T.: Quantifying the causes of path inflation. In: ACM SIGCOMM, pp. 113–124 (2003)Google Scholar
  37. 37.
    Streibelt, F., Böttger, J., Chatzis, N., Smaragdakis, G., Feldman, A.: Exploring EDNS-client-subnet adopters in your free time. In: ACM IMC, pp. 305–312 (2013)Google Scholar
  38. 38.
    Toonk, A.: How OpenDNS achieves high availability with anycast routing (2013).
  39. 39.
    Woolf, S., Conrad, D.: Requirements for a Mechanism Identifying a Name Server Instance. RFC 4892 (2007)Google Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  • Ricardo de Oliveira Schmidt
    • 1
    Email author
  • John Heidemann
    • 2
  • Jan Harm Kuipers
    • 1
  1. 1.University of TwenteEnschedeThe Netherlands
  2. 2.USC/Information Sciences InstituteMarina Del ReyUSA

Personalised recommendations