Ce contenu est disponible en anglais seulement.
A few weeks ago, the Canadian Internet Registration Authority (CIRA) released the initial findings of our Internet Performance Test. Our test shows Internet speed beyond your ISP's network and also other indicators such as quality (through ping and jitter) and future-readiness (such as DNSSEC and IPv6 adoption).
For TV lovers out there, the data might also point to some red flags for anyone looking to purchase a 4K TV. The speeds reported by most Internet users in the country are not sufficient for satisfying 4K TV streaming from any available source on the Internet.
Next-generation TV experiences are coming (or are already here?)
It is no surprise that the television market is highly competitive and the manufacturers are looking for ways to keep consumers buying the next big thing. 3D TVs didn't make a blip and smart TVs haven't been the demand driver that many expected. The industry is looking to ultra-high definition content to convince the masses that it's time to upgrade their old sets. And it is working!
According to a recent HIS report overall TV shipments are in decline but 4K continues to show strong growth rising 173% year-over-year. In fact, 4K TV accounted for 50%of shipments of 55-inch-and-larger TVs in 2015.
The three ways to get 4K content is through the latest Blu-ray players, via download, or by streaming. The early indicators are that Blue-ray will not likely see a big spike since sales of discs continued to fall (by 12%) in 2015 (US numbers ). Whether or not 4K will be enough to bring customers back to the showrooms for physical media is a question for others to debate, but I think it's fair to bet that streaming is here to stay as the preferred way to get content.
What does it take for Canadians to turn on a 4K TV?
The average Canadian consumer will have a few obstacles to overcome. The first challenge is data caps. To quote a recent TechDirt article on the subject, “Netflix's House of Cards in 4K will eat broadband caps like popcorn shrimp”. Canadians buying a fancy new TV need to be looking at those data caps and thinking hard about their Internet service and data plan, and whether they need to upgrade.
The second challenge will be bandwidth. Bandwidth is (basically) a measure of the raw speed irrespective of quality. Speed is something the average consumer has not had to pay that much attention to in the last couple years. Everything we did on the Internet was done at lower resolution and with lightweight HTML. HTML 5 has come along to help make our experiences prettier, mobile friendly and keep things light(ish) weight so speed still didn't matter much. For the most part, for urban Canadians the available Internet speed has not been a problem.
Streaming video in ultra-high definition is about to change that.
Netflix recommends 25 Mbps connections for an optimal experience, while their CEO was once quoted that it would be optimal at 50 Mbps. Others have quoted that 25-50 MBps sustained speeds are necessary for an optimal experience.
Let's focus on that 25 Mbps minimum recommended speed. 25 Mbps may not seem so bad, but CIRA's analysis is showing that most Canadians will be quite disappointed with their experience. The average speed that Canadians are really getting on the Internet is only 18.6 Mbps –a number that surprised us because we thought it would be much higher. If a typical family has 2-3 people using the Internet at once for bandwidth intensive activities then you can forget watching that latest Pixar movie in 4K because you may just spend half your time buffering. This means that Canadians who took our test either don't have a faster package available to them or that they are choosing to not pay for a faster one. Let's be clear. The CIRA Internet Performance Test uses testing nodes located in Internet Exchange Points on the Canadian Internet and not necessarily your fastest and closest server. It is intended to be neutral and provide a fair average of your speed regardless of what site you are visiting. If your Internet connection is close to a server that delivers 4K content you should see better results from that server. Importantly, for the majority of Canadians we know that the ISP networks have sufficient capacity to deliver Internet services that meet the 4K minimums... but there might be a catch.
The elephant in the room – net neutrality
As defined by Wikipedia net neutrality is, “the principle that Internet service providers and governments should treat all data on the Internet the same, not discriminating or charging differentially by user, content, site, platform, application, type of attached equipment, or mode of communication.”
However, this is only a guiding principle that in reality is impacted by the architecture of the Internet itself. While we may strive for content being delivered from a distant home server to arrive at the same speed as that delivered from your local government's website, this principle isn't manifested in the real world. The Internet is a very complex network-of-networks and the reality is (however much we don't like it) some content is treated differently than others.
Let's look at an example that is sure to be one of the more controversial points for Internet pundits: the general ubiquity of 4K content available to subscribers of the big telecoms. Importantly, this capability is not delivered as part of their “Internet” experience but as part of their TV subscriber programs. They also have other controversial content distribution programs where an incumbent streaming TV service doesn't count towards data caps when used through the set top box. For many, this is a win-win because the content is available at any time for a good price and the telecom company gets to keep a customer. However, does it potentially present a non-level playing field when the content gets there faster and doesn't count towards your cap versus that delivered by an upstart tech company or independent film distributor.
Could peering offer more neutral content delivery?
So if your ISP is able to deliver 4K video really fast because it is stored and served on their own backbone then maybe you don't need to upgrade your Internet access. Maybe the 25 Mbps package you are paying for today is good enough for web surfing. In this scenario, big content providers with excellent global footprints and well connected to big ISPs (think Netflix and Microsoft) will be fine in delivering positive customer experiences. This means that they can get their content closer to Internet users with less concern for costly Internet transit. It is one of the reasons CIRA helped to support a network of Internet Exchange Points across Canada –so that everyone has a chance to connect and serve Canada in the most neutral way possible.
So how the heck does a consumer understand speed in this confusing new Internet world?
The first thing you can do is understand that the speed you are getting when accessing the entire Internet is different than your raw maximum throughput. Various large organizations and open source communities have offered various tests to do this –including the big players like Ookla who license their platform to ISPs and Netflix who is, presumably, trying to help subscribers choose a good ISP.
CIRA runs a test too and it runs from one of three Internet Exchange Points in Canada. All of these tests are 100% relevant but measure different things based on their objectives. CIRA believes that our method is the most neutral because it focuses on non-profit Internet Exchanges inside Canada where every major content provider, hosting company and ISP should peer. If they all peer it helps to make Canada's piece of the Internet stronger.
So I probably need to upgrade my Internet service. What else can I do to help make a better Canadian Internet?
For the reader, you can help us to collect even better data and produce even better analysis of the Canadian Internet if you take the time to run a few tests. Set up auto-testing and have it running in the background. We use this data to help identify technical issues like slow points on the Internet, but also make it available to researchers and municipalities who are interested in correlating things like Internet speed with other demographic data.
We were surprised by how “not ready” Canadian's appear to be
When we initially designed the user interface for the CIRA Internet Performance Test, we had assumed that the average Canadian was going to get around 25 Mbps. We assigned below 25 Mbps with the color red on our Canadian map and 25 Mbps a nice neutral white. Because real world performance has been much lower than we expected, what we now see is that the map of Internet speeds shows Canada is awash in red – which is to mean slower than expected. The average speed that Canadians are paying for/getting on the whole Canadian Internet (i.e. to Canadian IXPs) is only 18.6 Mbps. More importantly there is a significant urban-rural divide with rural households averaging 14.81 Mbps. Both of these numbers are well short of 4K streaming recommendations unless you are using your service providers 4K services that are delivered using a separate “piece” of the bandwidth pie.
What can you do?
Well, part of the solution is to ensure that you are getting great service on a nice fast connection. For the large number of Canadians that don't pay close attention to their service plans it may be a good chance to review what you are paying for and to look at the competing plans available in your region. The other part is more about what ISPs, hosting companies, governments and content providers can do. Strengthening Canada's Internet requires big fat transit in our borders (think super-highways of the Internet), but is also should include a lot more local peering (think distribution hubs of the Internet). This means that the speed that content is accessed and downloaded can be faster because it doesn't have to take a complicated route over the messy “Internet”. It also helps create a neutral environment where everyone can play – including ISPs big and small.
And finally –take a CIRA Internet performance Test. Tell all your friends to take one. Set the test up to run automatically over a period and see how your performance changes. It is only through collecting data like this that we can write these conclusions.