What criteria should be taken into account when calculating hosting for a site with a video contest?

Question

Video

The site will have a page with a video contest (voting from social networks)
Participants will upload videos to the site.

You need to figure out the hardware and how much it will cost:
1) Renting a server (and what is better, a dedicated iron server, VDS)
2) Buying a server

You need to understand what criteria and how to calculate the load, taking into account the features associated with video.
Maybe you need more than one server?
What kind of load can you expect with such a server?
2 x Intel Xeon 5130 2 GHz / 8GB DDRII / 4 x 160GB

SATA
To save on hosting, there is an idea to beat this in the form of users uploading to YouTube and adding a link to the site that expands into a frame. What are the pros and cons compared to the option of directly uploading your own server?

Answer 1

if posting a video on youtube, then the most common vds is enough for you.
and if the video is uploaded to your hosting, then there is already video processing by your server and a lot of traffic ... and a lot of additional gestures.
The most costly in terms of resources will be converting video from the downloaded format to web formats of different quality. Also the channel width. For example, for 100 people watching a video in HD quality.
so these are completely 2 different approaches, both in terms of cost and implementation.
I would advise you to start with a cheap vds + youtube upload and that's it.
statistics:
taken from youtube as a sample took www.youtube.com/watch?v=I41CX0RdmPQ
video in 1080HD quality for 1 person up to 15Mbps at the peak
of loading @DmitriyEntelisrightly noted that if we increase the channel to 1Tb / s, then we will already run into screws and so on ....
therefore, either only low quality on our site, or give everything to youtube

Answer 2

I'll supplement @Dimitriys ' answer with a small common sense:
How to evaluate such things in general:
You need to carefully write out all the functionality that takes place in the service.
In your case, this is video download, resize, storage, viewing, likes, etc.
Estimate the average and peak number of events of each type per second/hour/day/month.
Estimate the load on the processor, network, hard disk, memory for 1 event of each type.
Estimate the planned lifetime of this architecture
Multiply the numbers and get objective resource requirements.
Make an assessment by the number of events at least - then it will be possible to advise you something.