With a video sharing service, you have large amounts of data transfer, and data storage. You're basically pushing large BLObs around your servers and to your us

With a video sharing service, you have large amounts of data transfer, and data storage. You're basically pushing large BLObs around your servers and to your users. If you have a relatively simple page, you could get by with low compute - you're just transferring HTML, CSS, and the binary. HTML5 video players are simple and easy to find, I've done it in just a few lines for a demo. I actually had more problems converting the video than getting the HTML5 player to work. What resolution would that 3 minutes of video be? That might impact your plans. And how many people do you expect to see a single 3 minute video? Multiply the video size (in MB) by the number of expected downloads to get an idea of the bandwidth for that video. Multiply that bandwidth by the number of videos to get an idea for how much bandwidth you'll need for the whole site. Of course page loads and overhead will add to that, I'm not sure exactly how much but I'd add 20% to your bandwidth estimates just so you have some spare bandwidth. You should absolutely look into CDNs. They reduce the load on your servers while getting stuff to your users faster. I think Cloudflare might have a free tier, and other CDNs might too. Azure isn't as well-known as Amazon; if you're expecting to be able to Google blog posts when you have problems, I'd go with Amazon. And there are other cloud hosts, ranging from nearly Azure size, to much smaller. You might choose a lower-cost more-distant datacenter, and a low-cost or free CDN, to start your MVP. If you're interested in using .NET or IIS, then look into Bizspark.