1. Why isn’t the South African Internet liberalised yet? William Stucke SAFNOG 2 Swaziland 7 th – 8 th April 2015

2. Focus Areas ISP Association Licences Local Peering International Connectivity National backbone network Metro networks Last Mile networks A glimpse at Africa © William Stucke 20152

3. ISP Association Why is this important? Gives ISPs a single coherent voice when speaking to Government or Regulator Lends power to your arguments Avoids single ISPs from being “punished” for speaking out. ISPA formed in 1996 in South Africa Now has some 160 members, and represents the majority of active licensees. As a more mature market, also has: WAPA SACF Wi-Fi Forum © William Stucke 20153

4. Licensing Regime RSA had a very centralised market Only 11 “previously advantaged licensees” Vertically integrated Government owned a share in 8 out of 11 operators Others were all VANS – Value Added Network Services Telecommunications Act gave the Minister of Communications the right to make seven “Declarations” In August 2004, she announced five of these, which were properly gazetted and came into effect early in 2005 In late January 2005, days before they were due to take effect, she published a Press Release saying, paraphrased: “In order to clarify, I didn’t mean it.” Although not legally binding, she obstructed ICASA until they withdrew “Self Provision” for VANS from the new regulations. © William Stucke 20154

5. Licensing Regime 2 New Electronic Communications Act took effect in 2006 Diametrically opposed to the previous Act. Instead of protecting Telkom’s monopoly, it fosters competition. She instructed ICASA to determine if “a few, if any” VANS should get new licences under the new Act. August 2008 – Altech won court case against this. Appeals failed ICASA issued 451 VANS with licences on 19 th January 2009 Both iECS and iECNS licenses All (unless specifically requested otherwise) with full national scope Identical to the licences held by Telkom and other previously advantaged operators Except no spectrum Today, some 713 licences, including class and broadcasting licences. © William Stucke 20155

6. Local Peering JINX operational in 1996 Today, ISPA and NAPAfrica operate six IXPs: Joburg, Cape Town and Durban GINX (Grahamstown) closed in 2012 © William Stucke 20156

7. International Connectivity SAT3 /SAFE 2001 800 Gbps Managed by Telkom Expensive. High Cost, Low Volume model SEACOM 2009 1280 Gbps ZA, MZ, KE, TZ and UG High Volume, Low Cost model. 35% immediate price improvement for Medium ISPs, growing dramatically after 3 years, when capital purchase paid off. Cheaper to peer in London than buy local transit! Followed by EASSy 20104720 Gbps WACS 20115120 Gbps Resulted in dramatic reductions in National costs © William Stucke 20157

8. National Backbone Networks Now the fun starts! 2009:R10,000 / Mbps / month for national transit 2014: ~R1,000 / Mbps / month for national transit Multiple operators building competing networks Huge bureaucratic obstacles: Way Leaves, EIAs, etc. FibreCo took 2 years to get 19,000 permits for 1,600 km of cable – Joburg to Cape Town Solution? Rapid Deployment Guidelines © William Stucke 20158

9. Rapid Deployment Guidelines Electronic Communications Act, effective in 2006, obliged the Minister of Communications to liaise with the Minister of Provincial and Local Government, the Minister of Land Affairs and the Minister of Environmental Affairs to draft these. Nothing happened Draft produced by others, but not yet finalised Now the Department of Communications has been split, and we have two Ministers arguing about who’s responsible for what 2014 Amendment of the Act changed the names of the Ministers, and reduced their role to drafting a Policy Made it ICASA’s problem to draft the actual Regulations Still no Policy © William Stucke 20159

10. National Backbone Networks - 2 Some 19,000 km of fibre required to link all towns in SA with a population >10,000, as per the 2001 census. Only Telkom has this, with 104,000 km of backbone fibre Despite the fibre being laid, we haven’t yet reached the “tipping point” where connectivity suddenly becomes cheap: The High Volume / Low Cost Model. Lack of courage? National networks being built by Neotel FibreCo (Cell C, Convergence Partners and Dimension Data) Broadband Infraco – doing nothing. 100% Government owned JV (Vodacom, MTN, Neotel) DFA Long-haul SANRAL © William Stucke 201510

11. National Backbone Networks - 3 Telkom104,000 km Neotel 21,000 km FibreCo 12,000 km Broadband Infraco 13,000 km JV 5,000 km DFA Longhaul Liquid Telecom © William Stucke 201511

12. Telkom – February 2012

13. © William Stucke 201513 Neotel - 2011

14. FibreCo 2015 © William Stucke 201514

15. Broadban d Infraco 2012 / 2015

16. Liquid Telecom

17. Metro Networks A bigger problem than national networks – more fibre needed. Joburg Metro has >9,000 km of tarred roads Roads being dug up multiple times Some of those rolling out metro networks include: The Metros themselves Telkom Neotel DFA MTN Vodacom © William Stucke 201517

18. DFA Gauteng Metro Fibre

19. Last Mile Networks Only Telkom has a substantial fixed line Access Network ~70,000 km of copper 3.6m out of 8m lines active – and dropping every year Most lines provisioned for ADSL 927,000 ADSL subscribers or 25% LLU? My most significant failure of my time at ICASA FTTH Wireless © William Stucke 201519

20. High Demand Spectrum ICASA has tried several times to assigned “High Demand” spectrum, via a competitive process, over the last decade. Obstructed by Government Spectrum is “High Demand” because: It’s allocated for PtMP in national plans, Harmonised with other countries in ITU Region User equipment manufactured More spectrum essential to: Facilitate higher throughput on mobile networks, e.g. LTE Provide cost-effective broadband in rural areas © William Stucke 201520

21. What High Demand Spectrum could be assigned? 310 MHz Nationally 28 MHz x 50 Districts © William Stucke 201521

22. Progress so far? ISP Association Licences Local Peering International Connectivity National backbone network Rapid Deployment Policy and Regulations Metro networks Last Mile networks LLU Spectrum © William Stucke 201522

23. African Situation Same issues apply: - ISP Associations Licences Local Peering International Connectivity National backbone network Metro networks Last Mile networks © William Stucke 201523

24. ISP Associations AfrISPA2001 – 2010 Provided training in ~20 countries Helped increase ISPAs in Africa from ~ 5 to ~20 Died after Brian, Eric and I left Do we still need it? © William Stucke 201524

25. Liberalisation About half African countries have a liberalised telecommunications, with multiple ISPs operating Cross-boarder connectivity still a problem No consistent mechanism in place to arrange access to “No Man’s Land” and cross-border licensing © William Stucke 201525

26. Local Peering AfrISPA, KTH, ISOC and NSRC all supported the formation of IXPs During the AfrISPA CATIA project, African IXP countries increased from 5 to 22 Some countries punt the concept of Regional Peering Preconditions: Adequate, cheap national fibre Cross-border connections An operating and significant IXP in the host country See 2006 paper: http://www.afrispa.org/dpages/PositionPapers/RegionalExchanges.pdf http://www.afrispa.org/dpages/PositionPapers/RegionalExchanges.pdf © William Stucke 201526

27. IXPs in Africa 2001 © William Stucke 201527

28. IXPs in Africa 2002 © William Stucke 201528

29. IXPs in Africa 2006 © William Stucke 201529

30. IXPs in Africa 2011 © William Stucke 201530

31. IXPs in Africa 2015 © William Stucke 201531

33. Why connect to another Country? Local traffic delivered more directly Peering Quality of Service Redundancy

34. Talk to me about … Updating ISPA data in Africa Updating IXP data in Africa Restarting AfrISPA? Consulting opportunities Becoming Country Representative for BoardPAC Drinking whiskey William@stucke.co.za +27 79 501 0850 +268 78 713 733 (During conference) © William Stucke 201534