[reggieband]

Price is pretty good, surprisingly. Looks like they take RTMP in and output m3u8 (probably fmp4 behind the scenes). 2 second latency is also not bad.

Larger streamers on Twitch can get ~20k concurrent viewers. A long streaming day would be 10 hours. They would also be streaming from NA so using their example pricing suggests:

Input: $2/hour * 10 hours = $20

Output: $0.15 * 10 * 20,000 = $30k

A worst case scenario of $30,020/day of streaming. Still seems financially unviable. I guess Twitch can stay in business for now.

EDIT:

Basic table

    Viewers Hours Min     Best     Worst
    20000   10    $7,520  $14,020  $30,020
     2000    6      $462     $852   $1,812
      500    4       $83     $148     $308
      200    2       $19      $32      $64

[PaulRobinson]

Until 5 years ago I was CTO of a company called Livestation where we handled live streaming of about 40 news broadcasters including Al Jazeera, BBC World, CNN International, etc.

In my stint we dealt with the Egyptian and Turkish uprisings. The relevant diasporas were super keen to see what was going on in all those countries, and often the only places you could see them online was with us (we had raw feeds in a few places).

There is no way we would have paid $752/hour for 20,000 users for 10 hours. We would have dropped the bit rate way, way down (which is fine for most content), and we'd agreed pretty decent bandwidth rates all round.

Depending on context, 2 second latency might not be worth it either - spend less, get 10 second latency and use HLS using an off the shelf CDN, and you're going to pull the price down even further. When Akamai can offer you a better deal than this, you have to wonder how good a deal this is. :-)

We did end up using RTMP a lot for a P2P live streaming product, but that never took off, and eventually the company folded. Fun times, though.

[reggieband]

> When Akamai can offer you a better deal than this, you have to wonder how good a deal this is. :-)

If you were doing the volume of traffic necessary to get the Akamai discount in the range you are suggesting then you are also going to be able to negotiate Amazon pricing in a similar way. It is unfair to compare theoretical bulk CDN pricing to non-discounted Amazon pricing. It would be like negatively comparing the published price of some other CDN to your private knowledge of deals you did with Akamai that probably included contractual commitments.

If you go by published CDN pricing (I use Fastly as an example in another post below) and assume you are pushing close to the maximum bitrate then the comparison is actually reasonably fair. Anything outside of that is speculation based on theoretical deals you might negotiate based on theoretical usages.

However, I would have no problem agreeing that building your own system to handle large scale would probably get to a lower $/GB. But then you have to build it, maintain it, monitor it, etc.

[PaulRobinson]

My problem with Amazon at the time, was there was no mechanism to start that conversation. Most CDNs have account managers where you can have a call and discuss a TB/month commit, and price accordingly. I looked for ways to reach out to AWS to do the same, and no dice: they have the Hollywood principle in place, which means you'll get your account manager after you have already committed to them as a platform.

I'm sure that I would have got a better deal if only I'd figured out how...

[mabbo]

Imagine you're a University that needs to, as quickly as possible, build a system to have streaming video for all your classes. Otherwise, you've got students and parents saying they want their money back because they paid for real classes, with Q&A and a live professor, and not pre-recorded videos.

200 viewers for 2 hours sound about right for a typical large University lecture. Typical students are paying hundreds or even thousands of dollars per lecture. $64 per class would put a dent into the University's budget, but not a huge one. And the fact that you can now scale a class to 1000 to 2000 students simultaneously, well that's cheaper than hiring more professors.

In short, I don't think this is meant to compete with Twitch or Youtube. That doesn't mean it doesn't have a niche it can work within.

(Bias note: I work for a totally unrelated branch of AWS. My thoughts and words are my own and are not representing my employer in any way.)

[gundmc]

My University had GSuite for all students and faculty. Meet supports up to 250 participants for a normal call and Livestream only for up to 100,000 viewers.

I imagine Microsoft has a similar offering for schools that use their products. I don't see a case where it makes sense for schools to be rolling their own livestream platform and paying these fees.

Edit: I'm not positive if the Livestream capabilities are included in the free Education package or just the enterprise package.

[johnbatch]

Microsoft has live events that support 10k participants: https://docs.microsoft.com/en-us/microsoftteams/teams-live-e...

[wrkronmiller]

Interesting point. I'd be surprised if most universities were trying to build out apps for the current quarantine though.

More likely they just email links to Zoom conferences or something to that effect, no?

[asutekku]

That’s how it is handled in most universities in finland as far as i know. Lectures are just zoom meetings and everyone already knows how to use it.

[shams93]

Then there's also online proctoring for test taking for universities as well where you need to stream live test takers to proctors.

[yomly]

What? People go on camera to sit exams so they can be invigilated?

That's so short sighted. Anyone who wanted to cheat could do anything to circumvent that - first comes to mind is plastering the wall with notes...

[grogenaut]

When I did my ham radio test they had me show the camera around the room, desk, wall, ceiling, floor, anything that was easy to glance at. Or better just use my cell phone as a second camera looking at me from the back left.

I do find it funny that people get mad about this "invasion of privacy" except this is exactly what a live in person proctor is looking for.

[ALittleLight]

I don't get how it can cost so much, and also how YouTube or Twitch would let you stream for free. They can't be making that much money off a stream, can they?

A 10 hour stream viewed by 20,000 people... My recollection of YouTube is that they pay the creator approximately a dollar per thousand views, if that's a third of what they make, then they make 3 dollars per thousand views. Assume YouTube counts a view as ten minutes, so 200k hours of streaming would be 1,200,000 views (200k hours to minutes / 10) would be (120*3) 3,600 dollars?

Something seems off by an order of magnitude. Is the markup that high? Did I make a calculating mistake?

[reggieband]

You can't compare it to how much it costs YouTube. You have to compare it to how much it would take to build yourself.

Here is Fastly CDN pricing https://www.fastly.com/pricing which is a good enough indication of the market value of data transfer costs. Note that CDNs charge in $/GB (big B byte) and video is usually considered in Mbps (small b bit). Consider that the most costly tier of Amazon service ($0.17/hour) is based on a 8.5 Mbps 1080p stream. You should be able to use that to calculate if the service is competitive to bulk data transfer.

Quick math:

    8.5 Mbps / 8 = 1.0625 MB/s 
    * 60 * 60 = 3825 MB/h 
    / 1024 = 3.73 GB/h
    * 0.08 $/GB = $0.30/hour

So based on that math, not a bad deal. Of course, Amazon is assuming (rightly) that with adaptive bit rate most people won't be streaming 8.5 Mbps continuously. And since they own their own CDN (Cloudfront) they aren't paying $0.08/GB. In fact, anyone doing significant bandwidth would get a discount.

This is all very hand-wavy math but it should give you some insight as to why there are so few competitors in the consumer video space. Video delivery is just plain expensive from a data transfer perspective.

[notyourday]

It is worse than that.

In order to do streaming well (i.e. to prevent consumers from running away and coming back), one needs take an incoming stream and re-encode it in multiple bit rates - 1080p stream would have not only 8Mbit/sec but also 2Mbit/sec, and 128Kbit/sec qualities. In addition to that for a decent distribution there should also be a 4Mbit/sec and probably 64kbit/sec streams.

As players either hunt down or hunt up, one would be carrying a pretty big number of random cold chunks that are unlikely to be requested again at the edge, which has a negative impact on hit CDN hit rates.

[jorams]

I don't have a particularly good source for this, but I've seen this reported by major YouTubers: the advertising revenue split on YouTube is 45/55. YouTube keeps 45%, the creator gets 55%. So if the creator gets $1, YouTube keeps about $0.80.

[jpalomaki]

Price is likely set based more on how much customers are willing to pay for this service. The cost to provide the service can be much lower.

Target market could be for example companies that provide live streams as streams as service (equipment, people, production etc)

[goldenshale]

Because most recordings are never actually streamed to that many people. My guess is that the popular and expensive streams make profit while the rest cost basically nothing to host.

[joecot]

> I guess Twitch can stay in business for now.

I assume you're being sarcastic/coy, but for those that might not know, Amazon owns Twitch. I'm guessing this service is them white labeling Twitch's tech infrastructure, and priced such that someone could not make a Twitch competitor with reasonable profit margins.

[freehunter]

And they wonder why people complain about anti-trust.

[oldmanhorton]

Well, i don't think there are rules that they have to offer their proprietary tech at wholesale. I think it's a problem that Amazon is so massive as to own this tech at all, but once you have it, i would believe it's better to sell it for profit than to keep it locked up

[borramakot]

I've wondered for a while- are there heuristics on the income a steamer can make per viewer-hour?

Edit: I'd googled this before, but apparently I found the right way to ask: this site says from 1c to $1/hour/viewer. https://wallethacks.com/how-much-do-twitch-streamers-make/#:....

[cjhveal]

For what it's worth, that's just income directly from sponsored content. The bulk of livable streaming income (outside of huge endorsements) comes from monthly subscriptions, so conversion rate matters as much as total viewership.

[ponker]

This isn’t for consumer streaming — those people will still use Twitch. This is for people building things like corporate meeting software for which the users will never be close to 20,000 concurrent and the costs of $30k don’t even get looked at.

[hmmazoids]

Yup agreed -- looking at it from the lens of building a streaming website that scales to that many viewers per stream, it isn't going to be cheap (but cool it can handle it if you put the money in I guess).

I assume this is more targeted towards universities where they have a max of 300 students (at least that's what my university capped classes at) and they need something to get started quickly. At most they may stream for like 6 hours a day for each class. You don't even need good quality -- lowest tier quality (SD) ends up being

$2.00 per hour * 6 = $12 $0.0375 per hour * 6 * 500 = $67.50

Honestly ~$80-100 per day to stream to all your students? Still kinda expensive I guess but might be a small price to pay for a university if it means they can offer streaming education. Not to mention 99% of the time students don't even tune in and just watch the VOD.

[jontro]

Weird that they just not charge for egress as with other services.

Also it seems a lot cheaper, assuming full hd is 1 MB/s. That would mean approx 3.6GB/h which would come out at 0.54$/h in comparison to their 0.15$/h pricing

[reggieband]

> Weird that they just not charge for egress as with other services.

My guess is they are charging ingress in place of some equivalent of compute necessary to transmux. That cost is fixed regardless of number of viewers.

Their docs state "The maximum input for a Standard channel is 8.5 Mbps and 1080p resolution (full HD)." but they could be outputting something less than that.

[jontro]

The FAQ states the following "If you send the maximum 8.5Mbps, 1080p60 stream, Amazon IVS will create 8.5Mbps 1080p60, 3Mbps 720p60, 2Mbps 720p30, 1.2Mbps 480p30, 800Kbps 360p30, and 400Kbps 160p30 renditions in an ABR stream."

I assume you will only be charged for full hd if the client is streaming it

[mikeryan]

Cloudflare does $1 per 1000 minutes viewed at that pricing they're only $12k. (plus storage - likely only a few bucks more)

[atopuzov]

Twitch is owned by Amazon.