[sidhuko]

I built an app for major sports company in the UK several years ago now. The choice as actually the most cost effective. Most providers of the data, usually collated by people using an Xbox controller and relay device in the stadiums, are uploading static files to storage. We used a thin node.js layer to stitch these from a connected volume and expose them via an endpoint which would specify a time range. If a user never requested the results for the 36 min in a game then our solution wouldn't have to do any work. Once we did process a request for a user for a game at a given time range then all other users would then receive the same cached result. It was very cost effective when you compare millions of users requiring websockets and processing all the events regardless of whether someone is watching those results.

[brisance]

Thank you for your perspective. This is a perfect example of the difference between theory and practice.