12-16-05 | Blog Post
A CTO is frequently challenged with calculating infrastructure costs for new internet based initiatives. A favorite of small and medium sized business leaders is using the internet to stream audio and video. Whether for sales, support or training, product managers and executives properly suspect there’s something unique about using the Internet to deliver media content. CTO’s can help build a bandwidth cost model that can provide specific projections for various scenarios. It’s a matter of using some basic math against some very complex usage projections.
The amount of bandwidth in Megabits per second=SE x ER/1000
where SE=the number of Simultaneous End users viewing or listening to the media stream;
ER=the Encoding Rate of the media, generally expressed in kilobits per second (kbs)
ER determines quality of the audio or video. It’s the number of bits recorded for each second of media. The more bits, the better the quality (generally). Sound is generally encoded at 20 kbs up to 56 kbs. If you want to support 100 simultaneous listeners of an audio clip that’s sampled at 35 kbs you will require 100 x 35kps/1000 which equals 3.5 megabits per second (mbs) of bandwidth. If you only have 3.5 Mbs and a 101th person tries to listen they will get a server is busy signal and won’t hear the content. Video is generally encoded at 250kbs to over 512kbs. So to support the same 100 simultaneous end-users watching a 256kbs video would require 50 Mbs of bandwidth.
Since bandwidth is billed at the “95th” percentile and not solely on what you use during peak loads you’ll want to build a distribution curve of utilization. This curve projects the number of simultaneous users for the various times of the day.
By applying the basic math for bandwidth to the distribution of curve you can more accurately project bandwidth requirements and purchase only what you need.
To learn what OTC can offer for streaming visit OTC’s High Bandwidth webiste page
