117 subscriber-only videos ... around 150 views on average ... 815 for the most viewed.
so "bandwidth usage was within the top 1 percent of Vimeo users" says they have no audience and no traffic, doesn't it? The whole platform.
so "bandwidth usage was within the top 1 percent of Vimeo users" says they have no audience and no traffic, doesn't it? The whole platform.
Now, Vimeo's own delivery pricing is far below that -- perhaps an order of magnitude less -- but for most people there aren't going to be better pricing options for long-form video content of this sort.
That doesn't really contradict what I'm saying does it? Are you claiming that most subscribers watch most of the videos they watch? I'd be interested in seeing that stat for 3 hour videos!
Vimeo was either going to have to monetize video-minutes (which breaks their enterprise value prop and forces them to compete with YouTube and other streamers, which in turn requires a whole new set of corporate core competencies), or control costs on the low-end tiers by shedding low- or negative-margin customers. We're going to see a lot more of this moving forward, I imagine, as valuations start slumping.
I don't know what I would expect those numbers to be, but my instinct is that it seems like a really bad indicator for Vimeo's popularity. Either that or my instinct on how much bandwidth costs and how many views popular video sites get is way off, which could also be the case.
That doesn't surprise me. I imagine median customer as someone who created one video that 10 of her closest friends have watched. Anyone who has any following is immediately 95-th %-ile.
Slightly relevant post https://danluu.com/p95-skill/
I'm not sure which CDN. I looked at AWS EC2 and Fastly and they both cited outgoing bandwidth that was twice as high. Given that I can't find hosting providers with lower bandwidth costs, DO/Vultr either have very specific peering agreements or are subsidizing their bandwidth charges; either of which would be utterly broken by the amount of bandwidth that the person from the article appears to need for their multi-hour art tutorials.
Even if they had an amazing deal on transit, or were peered with everyone[0], and were able to provide $0.01/GB to video hosting at scale, we're still talking 20-40GB/view (assuming 45Mbps delivery of multi-hour video content). Delivering that over 800 views will run about $320 per video. Vimeo's cited custom plan pricing in the article is basically assuming a video per month on average at those rates.
[0] This is the particular reason why Google is able to provide YouTube at all. They are so big that every ISP absolutely has to peer with them, and that drives down their costs significantly.
it's probably using less than 10% of CPU or bandwidth most of the time
multiply it by 1000x and you can imagine how much leftover bandwidth there's left for the heavier users (beside they're not selling unlimited, you have a generous tier included and you pay extra for overages)
There's plenty of providers that can match the pricing
I've also worked for a company heavily relying on vultr and they have no problems handing high volume bandwidth (they're not as cheap as Hetzner!)
You'd also have to re-implement access controls and probably integrate a not-terrible video player. There might be open source software that does that, but nothing springs to my mind.
I've got the skills to do it, and $300/month seems high, but not unreasonable. That $300 a month frees me from doing updates, having to wake up at 3am because the site is down, having to transcode my own videos for lower-bandwidth clients, etc. It's probably roughly on par with paying someone else to do all of that stuff.
Works well enough for me as a slingbox alternative but I'm also not running it as a business.
[1]: https://hls-js.netlify.app/demo/ [2]: https://github.com/aler9/rtsp-simple-server
(Also, the economics shift radically if you're doing tens of petabytes a month; you're comfortably under $0.01 per GB at that tier, and the largest players -- major streamers and game companies -- can push their prices under $0.001. It's basically the folks who push terabytes who are getting hammered.)
This seems like a good area to get screwed, ie, pay a huge difference between a COTS black-box solution and just doing it myself. Last week I started playing around with setting up a RTMP server.
But hell, I'm still back to bandwidth. Are you saying there's a way to directly cache realtime streaming video on the cloud providers, skip the store-and-publish route entirely?
One thing I was amazed with was the huge amount of money a company could spend going into this area without doing some serious research. There are too many options and too many variables for most non-tech folks to consider.
No problem if you don't want to answer. I thought it might be something other HNers would want to know.
- they're unlikely to use public cloud providers (prohibitively expensive for high bandwidth use cases)
- they have more customers with similar needs to you
- its still a managed service, let them take care of it
the other comment already mentions Cloudflare, you could also consider https://bunny.net?ref=6akqfap0uq which is a small CDN provider focusing on cost effective solution (10x cheaper than AWS/GCP easily)
This seems incorrect. Azure changes at least 8 cent per gigabyte https://azure.microsoft.com/en-us/pricing/details/bandwidth/
According to [2] the 99th percentile of bandwidth usage "usually starts at 2-3 TB per month"
Van Baarle has 4,446 patrons [3] and assuming a 1080p video stream is 5 Mbps, to hit the 2TB threshold would only need 12 minutes of video per patron, per month.
And you're right, of course, that if you're paying $0.05/GB for outbound bandwidth and using 2-3 TB/month, you're using $1200-$1800 of bandwidth per year. So $3500 isn't entirely beyond belief.
Still, the takeaway from this, for me, is that Vimeo's "unlimited bandwidth" is anything but.
[1] https://vimeo.com/upgrade [2] https://vimeo.zendesk.com/hc/en-us/articles/360037761072-Ban... [3] https://www.patreon.com/loish
Depending on what your application is, this may never be an issue, or, if you're pushing heavy file and content traffic, it could form a huge chunk of variable costs that you need to figure out cost recovery for in your business model.
(All this gets upended if you're on something like SFDC or Dynamics, though, where costs looks nice and forecastable until you get to disk storage, where their pricing models appear to be put together by the kind of guys who usually present their invoices alongside a menacingly-brandished lead pipe.)
That would be surprising to most people I’d expect, totally not deterministic and kinda hostile. A more explicit “unlimited = 2TB/month” no matter your percentile would be more honest, but then you’ve said “unlimited == limited” which is a lie.
The percentile thing seems like a fairly obvious hack to keep using the word unlimited. Probably an MBAs idea..
Yeah, it seems like one thing that could happen as a result of this is that the current top 1% of users will leave the platform, creating a new top 1% from the people who used to be the top 2%. Then these people will probably leave the platform as well, creating a new top 1%, and so on. It's a nice algorithm for getting rid of your most successful creators.
step 2 of any analysis: project forward
You did this well.
https://vimeo.com/blog/post/improving-policy-on-video-bandwi...
This is extremely expensive compared to cloudflare stream. They price it by viewing time instead of by counting bandwidth and quotes ~$500 to serve 72,000 minutes of viewing time.
Or if you're on budget you can convert your video to HLS with ffmpeg and host it on your own server from providers with cheap bandwidth like hetzner or ovh to save on cdn cost.
So HillRat's $0.05/GB figure isn't the most expensive out there :)
[1] https://aws.amazon.com/s3/pricing/ [2] https://aws.amazon.com/cloudfront/pricing/ [3] https://www.keycdn.com/pricing
That is a ridiculous cost for bandwidth. It can only be meant for "low-volume" users or those that just don't care for some reason.
A platform like Vimeo should either negotiate a better deal, or start engineering their own CDN. After all content delivery is the main challenge for video platforms - everything else is boring and can be whipped together by a fund-less bored 15 year old on a weekend. If you're contracting content delivery out that's already kind of weird and pretty much guarantees everyone but yourself will make money with your platform.
If that's what's happening then this will wind up hurting vimeo's most loyal users.
We're comparing the wrong numbers. If it was a 1 minute video, probably not. If it was an hour long video, well, there's quite a difference in bandwidth consumed per view. The isolated view count is not a good metric for comparing bandwidth, unless you multiply it by the length of the video.
1000 people who watch a 1 minute video is 1000 minutes. 1000 people who watch an hour long video is 60,000 minutes, consuming much more bandwidth with the same view count.
What counts as a view? If someone watches 10 minutes of a 2 hour long video, does that count as "1 view of the entire video" for billing purposes?
- Vimeo is very wrong about how they calculate their bandwidth costs per user
- Vimeo doesn't have the scale necessary for peering and is thus getting a raw deal on bandwidth that you've managed to avoid somehow
- Vimeo is actually correct and you're about to become the next niche online creator subsidy vehicle
I'm not sure Vimeo is miscalculating, at least in the specific case listed in the article. If the creator in question were to self-host their own video on AWS or Fastly, they'd be paying about what Vimeo wants to charge them. If you're uploading very long videos (multiple hours) and have a use case (art tutorials) that demands watching at 1080p or 4K, then even a few hundred views will breach 10TB+ of bandwidth per video.
I have no idea how you guys would be even breaking even, at $59/mo plus $1 for every 1k views over 50k. I don't know what your bandwidth arrangements are, and I suspect you are not at liberty to explain them. But Vimeo is almost certainly larger than you, so they almost certainly have better bandwidth deals than you.
Of course, this one case could also just be an outlier, and perhaps Vimeo is hosing everyone else with high-margin custom plans. But I doubt this is the case.
[0] IP transit is typically charged by bytes sent, and peering costs are typically some large fixed cost to get a pipe of a specific capacity to a given network. A "view" can be anywhere from 1 to 50 Mbps depending on quality, and if you're paying for transit then your costs are also proportional to the length of that view.
The answer as to how is "deep tech" which we have been very protective of given how game changing it is. Not a satisfying answer I am guessing since it leaves out any hint of the how. Unfortunately necessary while we took care of protecting the intellectual property we created. Good news is that it looks like we will be able to share more in the near future.
Which is really sad, as they've generally been the only nominally-competitive youtube alternative.
If the US weren't so allowing of anti-trust behavior from big tech companies we might live in a world with several major video streaming sites that actually had to compete for creators and viewers.
Google gets to run YouTube because they can afford to store a lot of video, and every ISP needs to peer with them just to reduce their own bandwidth costs. Few, if any other companies have that advantage. The monopoly in online video is almost certainly being caused by upstream monopolies in last-mile Internet service, which makes them able to decide what companies can and cannot economically run an online video service.
The storage costs have to be astronomical as well. The latest stat I can find is that 500 hours of video are uploaded to YouTube every minute (~720k hours per day). Even assuming it's all standard definition, that's ~260MB per hour of video, which comes out to ~130GB of data uploaded per minute or roughly 187.2 TB per day. Full HD is roughly 5 times the size, so nearly a petabyte per day.
People were estimating YouTube uses an exabyte of storage about 6 years ago.
There are very few companies with the economies of scale to support running YouTube the way it currently is.