Cloud computing is a technology that has changed the IT industry and the way that companies operate over the past few years. At its core, cloud computing is the old “rent vs. buy” concept applied to computers, including servers and storage. The idea is that you can keep your total costs low by renting equipment that is hosted and managed by someone else off-site (“in the cloud”). Businesses are increasingly renting cloud-hosted servers and storage, instead of buying and physically owning/managing them (as they used to before). Just like the electricity in your home, cloud-hosted systems represent metered, on-demand, pay-as-you-go computing power and storage. Essentially, you rent as many servers or storage as you want, whenever you want, and you only pay for the time that you rent them. Forecasts estimate that cloud-related tech spending by businesses will be $235.1B in 2017 and that cloud applications will account for 90% of all mobile traffic by 2018.
Cloud computing is particularly suited for applications that experience peak periods of usage, interspersed with off-peak periods of usage. As in …. sports.
Cloud computing has definitely become a contender for the IT infrastructure powering the sports industry. For instance, UEFA’s cloud-hosted video service for Euro 2012 delivered 3000 on-demand, in-match and post-match video clips to subscribers worldwide, a total of 27 terabytes of HD content in all.
Over the last 3 years, it has made a lot of financial and operational sense for us to move our IT infrastructure to the cloud at YinzCam, given the nature of what we do for our clients–sports teams, leagues and venues.
1. The sports fan’s consumption of content is ideally suited to a cloud infrastructure. Take a mobile app for an NFL club (we support 26 out of 32 NFL club apps at this point). While the app will have a lot of news, photos, player interviews, and other media being published all 6 days of the week, the usage on Sunday really stands out, and justifiably so. Fans tend to consume the most information pre-game (injury reports, matchups, tweets, game preview), in-game (box scores, team stats, player stats, scores), and post-game (press conferences, blogs, game recap) on Sunday. The traffic seen by an NFL club app on Sunday during the season can easily exceed the usage on the other 6 days. The typical traffic/usage pattern looks like the graph below, with the Sunday peak traffic/usage often being 30x or more (sometimes, 100x for high-profile games) of those on the other days of the week.
Say that we hypothetically need 100 servers to support that significant Sunday peak traffic for this one app; those 100 servers will clearly be mostly idle on the other 6 (non-peak) days of the week. It makes little practical sense for us to buy and manage servers that would sit idle for most of the week/year. In addition, we would have to hire people to manage a LOT of servers, we would have to pay for the cooling/infrastructure/space costs associated with these servers, etc.
Instead, we can effectively support the traffic profile shown above by leasing those 100 servers, from a cloud provider, only for the duration of the game-day alone, and by leasing only a handful of servers on the other 6 non-game-days of the week. By using this lease vs. buy strategy for our servers and storage, we can also manage our total cost of ownership by reducing the number of IT staff needed around the clock to administer the servers. This has kept our costs down.
2. Time-to-market and time-to-deploy are dramatically shorter. When we sign up a new client, we don’t need to go through the process of buying/shipping/installing servers or storage for them or for us. With a few clicks on a dashboard, we simply rent a few servers on the cloud, and that client’s mobile app is up and running. This allowed us to deploy clients’ mobile applications faster.
3. A cloud-hosted system can be scaled up/down quickly. While game-day obviously displays the most significant load, there are the inevitable moments of interest that arise that lead to more usage, more content, and sudden relevant spikes of traffic from fans, such as the well-reported “Tiger effect,” when the 2008 U.S. Open playoff between Tiger Woods and Rocco Mediate resulted in a 15-25% increase in Internet traffic, with the peak traffic during play at the 19th playoff hole. Again, with most cloud platforms, with a few clicks, you can literally add unlimited storage and an unlimited number of servers, in minutes. This allows us to handle significant spikes in user traffic, from the sudden surge of fans wanting to watch the live announcement of a player’s retirement on a Monday afternoon, to the live streaming of a number of “war-room” video cameras from teams throughout the NFL Draft.
Moving YinzCam’s IT infrastructure to Amazon Web Services has allowed us to grow faster and more strategically, as a business.
- We were able to deploy the mobile apps for our clients faster.
- We could dynamically scale to support sudden increases in load.
- We avoided the equipment and people costs of buying and managing our own servers and storage.
- We were able to support teams/venues outside the United States simply by making a copy of our software to run in data centers in those regions (as seen in the map of Amazon’s global cloud infrastructure below).
Using cloud computing, we have successfully handled the extremely high usage that accompanies the NFL Draft, the Super Bowl, the NBA playoffs, the Stanley Cup playoffs, and the NRL Auckland Nines matches. We have scaled our system seamlessly to support the mobile applications of 85+ sports franchises over 5 years, with some 1.326 petabytes of content having been served up to 16M+ unique devices across the world.