A. . (s mw For more information on this report or to request data for your own region, please contact CIRA at info@cira.ca or 1-877-860-1411. © Copyright 2016 Canadian Internet Registration Authority (CIRA). All rights reserved. This material may not be reproduced in any form without written permission. The average upload and download speeds specified in this report have been calculated by averaging the results achieved by users of CIRA’s .CA Internet Performance Test in the applicable region over the applicable time period. Test results and statistical data validation was completed by Daisy Intelligence Corporation. Based on these results, CIRA has made various conclusions in the report. Notwithstanding, upload and download speeds can be affected by many factors, including, the Internet Service Provider (ISP) being used, the package purchased by the user, and network traffic dynamics (such as service availability, the number of users accessing the ISP’s network, and the time of day when the user is accessing the ISP’s network). CIRA does not represent that the average upload and download speeds in each region in this report is reflective of all users of that region. Any reliance upon the information in this report is at your sole risk, and CIRA shall not have any responsibility or liability for your reliance on this report. 2 INTRODUCTION The Canadian Internet Registration Authority (CIRA) manages the .CA domain on behalf of all Canadians and works to build a better online Canada with products, services and tools that support Canada’s Internet community. As part of these efforts, CIRA has built the CIRA Internet Performance Test with test nodes across Canada. To date, the test has collected data from more than 140,000 tests from individual users across the country. CIRA is indebted in this communal commitment to building a better Canadian Internet. CIRA’s test continues to collect data and reports and data will continue to be released at regular intervals. This report is made up of 126,000 tests collected between May and December 2015. Why measure Internet performance? There is a growing national consensus in Canada that access to fast, reliable and affordable Internet service is not only critical to economic development, but also represents an important social good. Understanding how Canadian users experience services in their homes and offices is an important first step in improving Canada’s Internet performance. CIRA’s Internet Performance Test is designed with the public interest at its core. The test nodes are located in Internet exchange points, rather than within an ISP’s architecture. This allows CIRA to measure the actual performance of an Internet connection in real network conditions, closely representing the actual Internet experience of Canadian users. Take the test 3 PERFORMANCE IS MORE THAN SPEED Like a finely tuned automobile, the lap time around a track is about more than just the top speed. Cornering, acceleration, braking, and driver skill all contribute to a car’s result. On the Internet, raw top speed is important, but so are other measures like ping time, jitter, packet sequence, congestion, and more. In fact the CIRA Internet Performance Test measures dozens of different attributes about your connection and makes them available in the advanced view. “What we want to see is a better online Canada. Where Canada not only competes with, but tops the global leaders in access speed, quality, and data sovereignty.“ -Byron Holland, president and CEO, CIRA 4 CANADA’S OVERALL PERFORMANCE Average download speed: 18.64 Mbps Average upload speed: 7.26 Mbps Average round trip time for a data packet: 96.43 ms Average jitter: 304.66 ms IPv6 Support: 2.4 % DNSSEC Support: 14.93 % The Internet Performance Test data shows a national average download speed of 18.64 Mbps. To put this finding in context, it is well in excess of the 5 Mbps that Industry Canada has targeted for all Canadians. However, 18.64 Mbps is well below the 25 Mbps that some other widely-available speed test applications report. Why? Because the CIRA Internet Performance Test is based on performance of an Internet connection in real network conditions - including conjestion, complex traffic routes or other network dynamics. This is a slightly different way to measure the real world experience of Canadians when using the Canadian Internet. This is underscored by the Akami State of the Internet report that shows Canadians rank 21st globally with an average of only 11.9 Mbps. Similar to CIRA, Akami collects this data based on real world access to the content it hosts on its network - a perfect demonstration that there are multiple valid ways to test the Internet. Using another data point for global context, Canada ranks 13th out of 92 countries on the Berkman Center for Internet & Society’s Internet Monitor. This ranking is based on a number of factors like adoption, speed, and price. Notably, this report shows that broadband adoption among households connected to the Internet is 83%. So how is Canada performing? 5 COMPARING PROVINCES Canada’s provinces and territories did not perform equally. New Brunswick leads the way with an average download of 26.98 Mbps and an average upload speed of 12.39 Mbps. The Yukon Territory has the lowest average download speed at 6.68 Mbps. DOWNLOAD SPEEDS 25 – <27.5 Mbps 20 – <22.5 Mbps 17.5 – <20 Mbps 6.7 Mbps 15 – <17.5 Mbps 12.5 – <15 Mbps 15.4 Mbps 7.5 – <10 Mbps 19.7 Mbps 16.6 Mbps 13.5 Mbps 22.2 Mbps 12.6 Mbps 21.8 Mbps 5 – <7.5 Mbps 15.9 Mbps 14.6 Mbps 27 Mbps 21.7 Mbps *The download speed data for Nunavut and the Northwest Territories has been combined. It is important to note that in provinces and territories where a high concentration of tests originated in an urban center the data can over-estimate the experience of many in the entire region. This is of particular concern in the north. Yellowknife, where access is generally good, dominates the averages for the NWT + Nunavut combined. The experience of many in NWT and Nunavut will be well below the 15.4 Mbps reported here. 6 COMPARING OVERALL SCORES OF 25 CANADIAN CITIES CIRA analyzed municipal-level data to help create a picture of how Canadians experience the Internet across the country. To help understand the relative performance of Canadian cities, the following chart takes into account speed, quality, and futurereadiness to create a single score for Internet performance. Each factor is weighted based on its contribution to the overall Internet experience. Cities were scored based on whether they were above or below the Canadian average for each category. Speed metrics were given twice the weighting of the other two. Ranking of Canadian Cities’ Internet Access 1 Toronto 14 Markham 2 Ottawa 15 Gatineau 3 Montreal 16 Lonqueuil 4 Fredericton 17 Burnaby 5 Saint John 18 Charlottetown 6 Quebec City 19 Vancouver 7 London 20 Victoria 8 St John’s 21 Edmonton 9 Mississauga 22 Surrey 10 Brampton 23 Whitehorse 11 Calgary 24 Winnipeg 12 Hamilton 25 Regina 13 Halifax The following series of charts rank cities (provincial capitals + the larger cities) and compares their performance to national averages. 7 SPEED – A COMPARISON OF UPLOAD AND DOWNLOAD SPEEDS AGAINST THE NATIONAL AVERAGE It is interesting to note that the east is significantly represented in the top right as having users with fastest connections. Even more interesting is that three of the Atlantic capitals: Halifax, St John’s and Fredericton are represented. The only other two cities in Canada that exceed the national average are Ottawa and Toronto. There is no representation of municipalities west of the Ontario border represented as above average in download. Canada Average Download (18.64 Mbps) 16 Quebec City Fredericton 14 Halifax Saint John 12 St.John's Montreal Toronto Ottawa Upload (Mbps) 10 Burnaby 8 Canada Average Upload (7.26 Mbps) Vancouver Gatineau Edmonton 6 Hamilton Brampton London Charlottetown Longueuil Markham Calgary Surrey Regina 4 Mississauga Winnipeg Victoria Whitehorse 2 0 0 5 10 15 20 25 Download (Mbps) *Relative city population size is indicated by the size of the icon. 8 30 35 QUALITY – A COMPARISON OF PING (OR ROUND TRIP TIME) AND JITTER AGAINST THE NATIONAL AVERAGE A positive Internet experience requires more than simply upload and download speed. CIRA has worked to create a quality metric defined by ping and jitter. When municipalities are mapped by ping and jitter next to national averages a large grouping of the largest cities land in the top right quadrant. Canada Average Ping/RTT(-96 ms) 100 Longueuil 150 Saint John Gatineau 200 Victoria Vancouver London Markham Fredericton Burnaby Toronto Calgary Montreal Jitter (ms) 250 Edmonton 300 Quebec City Brampton WhitehorseHalifax Regina 350 Canada Average (304.66 ms) Ottawa Surrey Hamilton Mississauga St.John's 400 450 Winnipeg Charlottetown 500 -140 -120 -100 -80 -60 -40 Ping/RTT (ms) DEFINITIONS Ping: round trip time for a packet of information, lower number is better, can affect applications like online gaming Jitter: the variation in time that a series of packets arrive, a lower number is better, can impact services like video and VoIP 9 *Relative city population size is indicated by the size of the icon. -20 Larger cities are over-represented in the top right because when you run a test, the quality you get is impacted by the complexity of the route your data needs to travel to get to the M-Lab testing servers located in Internet Exchange Points. In the absence of great peering and transit across Canada, larger population centers have a closer path to travel and more robust networks. Poor quality Internet connections affect the performance of common applications like voice over IP, streaming video, or media downloads. 10 FUTURE READY – THE PERCENTAGE THAT HAVE ADOPTED THE LATEST INTERNET STANDARDS FOR DNSSEC AND IPV6 As Internet technologies evolve, there is a need to implement new standards. CIRA has long advocated for both IPv6 and DNSSEC adoption in Canada and the Internet performance test measures progress towards this goal. Canada Average (14.93%) 8 Longueuil 7 Gatineau 6 Whitehorse 5 Montreal IPV6 (%) Charlottetown Quebec City 4 Hamilton 3 Calgary Halifax Surrey 2 Edmonton Ottawa Burnaby Victoria Vancouver 1 Toronto Winnipeg London Brampton Mississauga Markham St.John's Saint John Regina Fredericton 0 0 5 10 15 20 DNSSEC (%) *Relative city population size is indicated by the size of the icon. 11 25 30 35 Canada Average (2.4%) The Internet Performance Test will continue to track these standards and report on progress. It is important to note that Canada lags behind its global counterparts in terms of standards adoption. For instance, IPv6 recently celebrated its 20th birthday by reaching 10% deployment globally with Canada’s average only reaching 2.4% (http://arstechnica. com/business/2016/01/ipv6-celebrates-its-20th-birthday-byreaching-10-percent-deployment/). The Internet is a perfect example of thinking globally and acting locally. We can all best participate by ensuring that our piece of it is at its best and understanding what happens to data after it enters “the cloud”. 12 COMPARING URBAN VS RURAL – RURAL COMMUNITIES ABOUT 25% BEHIND There is little doubt that the power of population plays a role in what ISPs deliver across Canada. For the most part, in the large metropolitan areas, ISPs have built the networks and are competing aggressively for customers. This is not always true in rural markets. CIRA’s test relies on a crowd-funded model. Rural homes that don’t have Internet access obviously cannot run an online Internet performance test. As test-volumes increase and small regions are better represented in the data it will be possible to draw more detailed conclusions. Urban vs Rural Download Upload Ping/RTT Jitter DNSSEC IPv6 Mbps Mbps ms ms % % Urban 19.80 7.66 87.90 287.08 15.10 2.43 Rural 14.81 5.96 124.36 358.6 14.39 2.30 13 In what is sure to be one of the most hotly debated points in this report, the data among those who have run tests, suggests that rural access is relatively good at 14.81 Mbps download speed. When comparing this to the national average and urban average, rural communities are about 25% slower than urban ones and have significantly lower quality connections for RTT and Jitter. CIRA has adopted the Statistics Canada definition for urban (population of at least 1,000 and a density of 400 or more people per square kilometer). This report includes CIRA 29,451 tests from rural areas and 96,511 from urban centres. Recommended for HD video streaming services: 5 Mbps Spotlight on Ottawa Even within a large metropolitan area you can see that there is an urban/suburban/rural difference in download vs upload speeds that Canadians are getting. *Sample size for Manotick and Greely is very small (between 100-200) Download (Mbps) Upload (Mbps) Ottawa 22.53 10.19 Orleans 17.95 3.66 Nepean 17.17 5.95 Kanata 18.31 6.02 Stittsville 12.87 4.27 Gloucester 20.07 10.08 7.1 1.83 6.08 1.3 Manotick Greely 14 Recommended for Ultra HD streaming services: 25 Mbps APPENDIX 1: FULL DATA BY CITY Listed by population size Quality Speed Population Future Ready Download Upload Ping/RTT Jitter DNSSEC IPv6 Mbps Mbps ms ms % % Toronto 2615060 22.33 10.63 72.62 262.44 33.82 1.76 Montreal 1620693 16.98 11.26 84.59 275.52 15.97 4.38 Calgary 1096833 14.72 3.73 94.61 265.32 6.91 2.50 Ottawa 883391 22.53 10.19 69.44 309.81 25.29 1.44 Edmonton 812201 12.32 5.65 99.44 299.19 4.68 1.76 Mississauga 713443 19.32 5.68 75.03 339.16 23.77 0.64 Winnipeg 663617 13.09 3.58 111.26 471.96 6.48 1.87 Vancouver 603502 16.36 6.83 68.78 210.03 7.36 1.41 Brampton 523911 22.50 6.56 72.68 306.67 18.24 0.75 Hamilton 519949 17.84 6.63 69.73 322.25 29.38 3.05 Quebec City 516622 17.67 15.78 72.11 291.07 8.70 3.85 Surrey 468251 16.86 3.87 66.10 330.28 5.20 1.73 Halifax 390096 22.91 12.92 106.12 344.27 3.69 2.24 London 366151 20.24 6.05 75.96 214.90 19.14 0.94 15 Quality Speed Population Future Ready Download Upload Ping/RTT Jitter DNSSEC IPv6 Mbps Mbps ms ms % % Markham 301709 18.99 5.49 63.49 217.10 33.82 0.29 Gatineau 265349 13.77 6.49 80.87 184.65 12.13 5.95 Longueuil 231409 16.15 5.78 65.68 139.61 6.89 7.49 Burnaby 223218 17.36 8.59 64.23 255.01 8.33 1.39 Regina 193100 8.73 4.25 135.32 357.47 2.79 0.15 St John’s 106172 24.37 11.76 93.83 395.3 2.56 0.18 Victoria 80017 16.33 3.32 66.84 177.61 6.23 1.18 Saint John 70063 24.56 12.69 83.70 169.09 3.70 0.22 Fredericton 56224 21.87 14.33 88.8 236.82 3.86 0.13 Charlottetown 34562 17.12 6.01 171.17 486.89 3.47 4.22 Whitehorse 23276 6.06 1.88 122.51 327.9 3.14 5.49 Population from the 2011 Canadian census In order to put the priority on household connections and to remove high speed outliers that could include individuals testing from work locations, universities, etc, a test with a download speed greater than the mean plus twice the standard deviation from the mean was excluded for this comparison. Only cities with 100 or more tests were included in the analysis. A full review of all the data is available to qualified individuals and organizations. 16 DATA BY PROVINCE Upload Download Ping/RTT Jitter DNSSEC IPv6 Number of tests Mbps Mbps ms ms % % n Alberta 4.95 13.46 115.72 288.65 8.37 2.06 12 404 British Columbia 5.19 16.65 85.98 274.96 7.89 1.91 19 010 Manitoba 5.22 12.62 144.77 436.45 11.56 1.93 4 151 New Brunswick 12.39 26.98 103.17 235.95 5.48 0.48 3 560 Newfoundland 8.29 19.72 148.68 379.86 11.39 0.13 1 589 Nova Scotia 9.10 21.73 150.7 496.14 9.41 1.94 7 170 Nunavut + NWT 6.02 15.44 108.14 127.82 5.93 9.20 641 Ontario 7.90 21.76 86.29 298.91 25.09 1.54 47 215 PEI 5.80 14.60 205.52 858.75 3.48 2.43 948 Quebec 8.05 15.87 79.57 263.19 10.25 5.00 26 280 Saskatchewan 6.28 22.22 117.98 238.53 4.35 0.50 2 599 Yukon 1.80 6.68 139.75 242.47 4.05 6.08 395 17 APPENDIX 2: HOW THE TEST WORKS It is important to understand how the CIRA Internet Performance Test works in order to fully interpret the numbers. There are two basic ways that the Internet is tested. The first is by driving as much traffic possible through the highest-capacity route available and seeing how quickly the packets arrive generally via multiple TCP connections. This is the Internet that Canadian consumers are paying for because an ISP can only guarantee the performance on their backbone. The second way to measure Internet performance is to measure how quickly the data travels from your PC or device to a server in the wild and back. This is a more accurate measurement of your real world experience when visiting web properties. The CIRA Internet Performance Test uses test nodes located in Internet Exchange Points (IXPs) in Canada to measure your performance on Canada’s piece of the Internet. What makes this test so important is that the anonymous data we gather measures the entire path your data travels to provide research into how to improve the network in Canada. 18 A CANADIAN APPLICATION BUILT ON A GLOBAL STANDARD The CIRA Internet Performance Test uses a global open-source platform called M-Lab in each of its testing nodes. M-Lab is behind important global testing initiatives for performance, transparency, and state. The front-end of the test is an HTML5 interface built by CIRA. This type of application allows for the maximum throughput for everyone but the fastest of connections, where the results can be limited by the ability/speed of the browser to execute the test code. We have deployed nodes in Calgary, Toronto, and Montreal and have given enough network capacity to meet the load 99% of the time. In peak periods following significant media coverage on important data announcements the nodes can become busy from time to time. 19 A CROWD-SOURCED TEST, CREATED IN THE PUBLIC INTEREST Crowd sourcing data has its advantages and disadvantages. One advantage is that you can quickly get a cross-Canada view and get testing from big cities down to tiny hamlets. Secondly, it doesn’t exclude those aren’t able to participate in scientific white-box types of tests like the CRTC ran in 20152016. The disadvantage is that there is a large segment of Canadians who have little to no awareness of the Internet and how it works, can’t describe with accuracy what packages they are paying for from their ISPs and have probably never run a test of their connection. This type of data sourcing would, by, definition exclude these test subjects. The average Canadian wants to know whether they are able to watch streaming video. This was underscored by a recent CRTC report that showed that 50% of Canadians polled had no idea what Internet speed they expected to get. How can you help? Run as many tests as you can. Tell others about it. Get them to run tests. The more data we have the more we can help to effect policy change on the Internet in Canada. 20