James Cridland's blog

A radio futurologist writing about what happens when radio and new platforms collide
This is an archive post from my old website. Not all links will work. For new posts, visit my main writing index.

« | Blog index | »

Live IP streaming hits a new UK record

Posted on Wednesday, June 23rd, 2010 at 8:09 pm. #

An improved counter

Rory Cellan-Jones tweets on his officially-sanctioned BBC twitter feed: “Breaking – BBC streams of England match peaked at 800,000 – breaks previous record of 350,000 set yesterday with budget,world cup, wimbledon”

That’s a really impressive figure: testament to Richard Cooper and his team at the BBC, who look after much of the digital distribution, and their partners at Akamai and Limelight. 800,000 concurrent viewers is a pretty massive figure for online streaming.

Is this an internet record?
No. Back in 2008, Google used Akamai to stream a live concert for them, which got a peak of 893,000 live concurrents, according to TechCrunch. And last year, the Obama inauguration speech got, according to Mashable, 1.3 million live concurrents.

Is this a UK internet record?

How did this compare against broadcast?
BARB reports that ITV’s figures for the USA v England match were a total of 12.3 million viewers. But:
1. This match was on a Saturday night, when most people were at home or in the pub
2. “Viewers” doesn’t equal “televisions turned on” – unlike a small computer screen, most people watch television in the company of others (whether family, friends, or drinkers).
3. As BARB helpfully say, “the programme audience is the average audience of all the minutes covered by the programme transmission” – this isn’t a peak, just a mean average. Given that the television coverage of the World Cup consists of mindless witter for 30 minutes, then the game, then more mindless witter, it’s certainly fair to think that the peak audience was much higher.

So: it’s difficult to compare this broadcast figure with a ‘peak streams’ figure. Don’t trust anyone who tries to do that, because they’d be insane. But. If we were to compare them, then you could argue that the online audience represented 6% of the total. (The ‘real’ figure could be twice that, or half that – almost impossible to guess).

How much bandwidth was used?
If you pop along to watch live BBC ONE via the BBC iPlayer, then the rather clever Embedded Media Player settles on a bitrate that your network can cope with. You can discover this bitrate and other technical details by right-clicking on the picture: when I did, I got 800kbps (96kbps) – which means the video element was 800kbps, and the audio element was 96kbps. I think that means a total of 896kbps for this stream. For me.

The difference between broadcast and IP is that if two people watch on their laptop in my home network, then that means my network has to cope with two sets of 896kbps streams. If four people watch, that’s four sets of 896kbps streams, or 3.5 meg. And if all 800,000 viewers were given the same bitrate, that meant a total of 716,800Mbps – or 716.8Gbps if you like it better that way. Wow. That’s one big internet pipe.

To achieve distribution for this on UStream’s corporate rates, you’d be looking at £134,000; while TBI Research gave some rough rates at the end of last year suggesting that the price could be between $3 and $5 per user for just 30 minutes – that’s £2.1 million.

Live serving it is one thing, but to my surprise, the bandwidth distribution cost for this isn’t too scary. Amazon CloudFront doesn’t support live streaming, but if you were able to serve the match on-demand on Amazon CloudFront, I calculate that you’d pay £9,732 in bandwidth costs.

What does this mean for radio?
First, it’s great to see that the internet is capable of such a high amount of live concurrent users of streaming media. The internet remains a great way of reaching at-work listeners: where it’s hard to smuggle in a radio, and reception conditions are far from ideal.

However, acros all platforms (including FM/AM), the total ‘live concurrent users’ for radio peaks at 17.6 million (between 08:00 and 08:30, weekday mornings, according to RAJAR q4/2008).

UK internet broadcasting has hit a new record – and has now reached 4.5% of the daily live concurrent listeners to the radio; many of those listeners in a mobile environment, too.

4.5% is a brilliant figure. But we’ve a way to go.


commenting at June 24th, 2010 at 1:56 pm

The 15 minute peak for the Slovenia v England game was 14.0m between 4.30 and 4.45pm. (I could look up minute by minute data, but that’s more effort :) That figure includes BBC One and BBC HD viewing but not out of home viewing.

James Cridland
commenting at June 24th, 2010 at 10:19 pm

Useful figure, Sally – thank you!

James Cridland
commenting at July 8th, 2010 at 12:08 am

Steve Green commented about this blog on alt.radio.digital; I felt I ought to respond:

On Jun 24, 8:52 pm, “DAB sounds worse than FM”
> http://james.cridland.net/blog/live-ip-streaming-hits-a-new-uk-record/
> After calculating that the BBC delivered 716.8 Gbps of streaming bandwidth
> for the last England game, he then made some conclusions for radio, which
> consisted of his usual anti-Internet radio slurs where anything he says
> that’s remotely positive about Internet radio is always accompanied by a
> caveat that slags off Internet radio and makes the case for using DAB:
> (snip)
> Cridland’s claim that only 4.5% of all radio listeners could be served via
> the Internet at present is of course either deliberately or incompetently
> ignoring the fact that (snip)

Hold on, Steve, I didn’t claim any of this at all.

In the summary of that blog post, I compare these two figures:

- 800,000 concurrent viewers to the live coverage of that England
match on the internet
- 17.6 million concurrent listeners to live radio every morning
(across all platforms)
800,000 is 4.5% of 17.6 million.

My blog post just points out that the *record* for concurrent internet
listening/viewing is just 4.5% of what broadcast radio delivers every
single day.

In that post, I make any “anti-Internet radio slurs”, I don’t “claim
that only 4.5% of all radio listeners could be served via the Internet
at present”, and I certainly don’t “make the case for using DAB” – I
don’t even mention it!

Cheers for the publicity anyhow,


This is an archive site, and comments are now closed.