@artbrock complemented the conversation explaining he doesn’t think Lambda is the best parallel. “It’s more like what it would take to host a content management framework, not just wordpress out of the box, but like your own custom framework for you community or business, because at the moment that’s what building a hApp is most akin to, I think.”, he explained.
@pauldaoust wrapped it agreeing that it’s like comparing apples to penguins. " It struck he that, in terms of billing, it was similar to Lambda in that it was pay-as-you-go for compute cycles.
So just for example, I pulled this from a post about pricing.
It depends on:
The application, and storage requirements of your App,
The amount of data your app transfers for an average user’s experience,
The scale of your application, and
Where you host it and on what hardware
For our company, a standard setup might be an AWS environment with:
EC2 Webserver
RDS Database server
CDN file server
Scenario
Let’s say your app has 500,000 users and 5,000,000 app opens per month.
Fixed Costs
Let’s say a single webserver, a single db server suffices (note: you will need a very well optimized backend to accommodate this scale on a single web/db server).
So, Fixed cost: ~$400/month
Variable Costs
Lets say that the average experience for a user to open the app requires 2MB of data transfer. Note: this number will range wildly between different types of apps.
AWS charges approx $0.09/GB transferred - which comes to .00018 per app open.
So for a month, 5,000,000 x $0.00018 = $900/month
Totals
Total cost per month: $1,300
Cost per user = $0.0026
Cost per app open = $0.00026
So based on the above 500,000 users and 5,000,000 app opens per month do the prices there seem normal? Article was from 2016.
Michael the maths seems right to me. Can you share the link to the post you are following?
Its one of many I have been looking at, the problem is, there are so many variables when it comes to Web hosting, include this, add this, leave that out etc.
What I am trying to work out (And I know it is only speculative and basic with limited data) is what the average application hosted for x amount of users pays for any of the big 3 hosting providers.
From that, I plan to calculate the average number of users, to try and determine a rough figure for what it costs for a medium sized application to be hosted, then do some hypothesis on a small percentage of that average number shifting to Holo hosting and from that work out based on an assumed Holofuel supply the cost comparison between both.
I am trying to get an idea of the savings that being hosted on Holo will provide, given there is little staff, no infrastructure, no back ups needed, etc
I would like to try and do the same for Storage as well.
Its all speculation but with some information regarding average prices of App hosting and Storage, and the billing structure of Holo hosts it will still be possible to start working out some early figures.
2 Likes
@lucas.tauil Can you have a look at this link I just found please?
Its a total cost of ownership calculator, does this take into account hosting costs the way you read it?
If it does, this is a very easy tool to use for comparisons.
1 Like
When live maybe this is something Holo can also incorporate on their website, a calculator to show total cost of ownership, Vs Colocation Vs Holo Hosting.
3 Likes
Gotcha! this dialogue between @pauldaoust and @artbrock might shed some light into the differences between the existing offers in the market and how Holo hosting intends to operate:
Paul asked details on how service logger and billing, will work. Particularly billing period, hApp publishers setting billing limits, compute vs storage billing and who are the parties in the service agreements.
Art explained It is not our intention for service logger to work like that. Holo doesn’t operate like a smart contract that says everyone will be paid within X minutes because the funds are already locked into the smart contract for release.
“So we don’t enforce billing payment cycles, we create a fluid marketplace and smart matching”, explains Art.
The gist is:
App providers set their requirements of hosts. Things like:
- uptime requirements
- thresholds for speed or storage
- max price they’re willing to pay
- required certifications or KYC properties (like located in Canada, certified for ecommerce, etc.)
Hosts set their requirements:
- Categories of apps their willing to host (and categories they want to block).
- Typical payment speed
- They can set per app pricing if they want (like I’ll host Wikipedia for free bug charge gambling sites a bunch)
Then chaos happens.
“The chaos I mean is that App Providers set up their methods of paying (likely automated) and checking for fraudulent billing. Maybe they check and pay every 5 minutes. Or every hour. Or every day, or week”, concludes Art.
Hosts don’t have to host apps that don’t pay quickly enough for them. But maybe some hosts are willing to charge a little more, take some risk, and get paid more later.
Regarding frequency of billing, Art explains that hosts select that too: “Service logger lets a host set a billing threshold , like once I’ve done 500 HOT of billing (or whatever approximates to a dollar or so), then you send a proof of service invoice. It also lets a host set a billing period , so if they’ve been providing a smattering of hosting to a low-traffic app that hasn’t hit their threshold yet, they can generate a bill each period”.
“Since hosts can set their own prices on compute / storage / bandwidth and App Providers can set their limits/thresholds/priorities, it creates a mutual marketplace and marketplaces have their own natural chaos” he concludes.
7 Likes
This is indeed very handy.
Thanks for taking the time to follow this up Lucas, appreciate it.
Looks like we are starting to see how this will operate. 
That’s great information.
I wonder if there will be a price matching type dashboard or algorithm in place to match the closest to both parties variables?
If there is an agreement in place between host and creator I wonder how there could be any fraudulent billing, or accidental over/under charging?
With HoloFuel both parties have to agree to the send/receive so I imagine that would take care of most issues before they arose.
[/quote]
It looks like a very good system, and I guess the beauty of it is, any larger corporations or business that wanted to host themselves would not have to worry about how the billing is as they are charging themselves, or hosting for free.
Which brings me back to something else you said about hosting earlier and mentioned that Hosts who self host will not need to pay the TX fees.
My understanding was that people can use Holochain for free without ever having to use HoloFuel, but if they do want to use Holofuel, which they need to, to self host, they still need to pay the TX fees to Holo.
Not sure if I read it or saw it but something from Art about your welcome to use it for free but don’t expect to be able to connect to https or something along those lines.
Either way its great to see what the system will entail and again because it is brand new, and not yet live its hard to compare it to anything, but I will come up with something. 
3 Likes
Michael, I’m more a weaver than a technical person, my skill is on connecting dots and learning as I do so. @thedavidmeister got into the game, he is curious if we are saying that there is no way to have a holoport serve up an index.html file.
"So we need to compile html/css/js and put it on s3 and then the only interaction with holo is via a javascript web client on top of a centralised static resources hosting?:, asks @thedavidmeister
2 Likes
More soon!
1 Like
@tel142 has put quite a bit of thinking on the subject. Here is a copy of his post on the Holo Forum:
"I have been pondering the pricing/charging basics for some time. Obviously, Art is an expert at currency as per his metacurrency project however I guess he is not willing to reveal his original calculations to the public. But I expect Art used some basic parameters to work out the prediction of 1USD token value. Don’t think this was plucked out of thin air but used when calculating the ICO token metrics plus the costs constraints/build for holoports. Each type of holoport must have a cost/capacity/earnings rate which was probably used during the design and then into production. Then when each holoport is online earning funds – they have the desired ROI and subsequent income.
So, here are some my initial thoughts:
- ROI of the holoports should be around 1year and then move into profit
- Based upon 360 days the payment per holoport should therefore be around 2USD to achieve ROI for the 500USD one
- So maybe 1 to 2 Holofuel tokens per day based upon Arts 1dollar estimate
- The ROI should include operating costs for electricity, ISP & management
- The earnings must deliver the ROI/Profit including covering the operating costs otherwise people wont run holochain (its not a charity)
- The reserves accounts will be used to regulate price to ensure ROI is always achieved to guarantee the necessary resources are available to the network.
- As demand increases –the price will increase above the ROI price to encourage more resources to join network even if it’s the same people buying more hardware. Price will reduce as necessary resource increases, but not below ROI (they dont want to lose resources and the price to yoyo).
- ROI will change through phases (phase 1=holoports, phase 2=Linux, phase 3=windows, phase 4=android)
- If there is more resource than demand the quantity of holofuel will be reduced hence driving up the price to achieve ROI rate for every phase
- At Phase 1– the game will be to provide ROI to holoport hosts plus a resilient network for Happs without too much focus upon competitive pricing with AWS/google
- When Android/Windows version (phase 4) is released and running smooth – only then will the network have the capacity to challenge AWS/Google etc and then the ROI reduces bcus hardware/ISP is not a cost bcus they are using spare capacity to run holo – only then the price needs to be lower/competitive against traditional cloud tec. Also – holoports owners should have recouped their initial capital investment so will be OK with the reduction bcus now they are in profit with just operating costs to be incurred.
Not sure these thoughts are of any help to your calculations – but just thought worth sharing with you))
Not related to your price project but as an output of my thoughts:
a) The team are probably working towards a holding pattern for release bcus there is zero/very little demand for holoport resources yet so they are waiting for developers to catch up.
b) If they release real holofuel with no demand - the price will plummet way below ROI and will cost a fortune to hold reserves to bring the price up.
c) The best time to sell will be just before any new hardware platform is introduced.
d) The roadmap doesn’t include the ports to windows or andriod so these may be released dependent on demand rather than completion of the code as they could initially reduce the price of holofuel bcus ROI is much lower.
e) Holoport owners will be encouraged to add more storage. This will be to give the network more ability to host apps like youtube fb which will need much more hard-disk storage than processing power. Think could be why holoports come with 4 USB – V3 for storage extension and V2 for keyboard/mouse whilst keeping production costs low as possible and minimising initial ROI.
I don’t think you will get much feedback from Holochain bcus their pricing needs to be a black box otherwise whales (tokens and resources) could manipulate the network or token metrics … which would destroy the principle of a stable currency to be used to pay bills, give a steady ROI, plus maybe impact the resilience of the network/resource level if a large resource is switch off and on".
4 Likes
Here is @Dave’s reply: " Wow. You’ve put a lot of thought into this! Thanks for posting this info.
Here are a couple things that come to mind for me…
There are ways we might be able to roughly estimate some things regarding HoloFuel dynamics. For example, app providers will probably want to understand their hosting costs.
As part of that, something potentially useful might be to…
- get a figure for the value of the current web hosting services marketplace
- make a wild guess as to the percentage of the market that Holo might have at a certain time
- we can probably get a number for the quantity of HF that exists at a given time
With this information we can do some calculations and maybe get a snapshot in time of one aspect of hosting related costs. Will this actually be useful? Will it give us the information we’re looking for? I have no idea.
Other things that I try to keep in mind…
When the transition to Beta occurs, cost for app providers compared to Ethereum will be less. Compared to AWS will be more.
Assuming equal or better performance than Ethereum, I would expect this to drive app providers to the Holo network.
The expectation (so it may not actually work this way) is that the price for Holo hosting will be driven down over time by market forces so that it eventually is on par with, or maybe even better than, other companies like AWS.
Which means that an app provider will be able to purchase an increasing volume of services with the same amount of HF over time. Until it gets to the point of being on par with other web hosting companies in the marketplace.
So if I’m an app provider, I can (possibly) get web hosting services at a discount right now (assuming things work they way they are expected to).
How much of a discount? Maybe finding the delta between cost at launch vs. cost in existing market? This will be different over time as the cost moves to a stable point.
I feel a need to say that I might be entirely wrong in my assessment.
I hope this is useful in some way."
1 Like
@tel142 reply to my post: “Backup and archive are always forgotten about in most apps so good to hear you guys are addressing this 
Hopefully along with other functions user can use like a good search function; stats like costs to store and projected costs; resilience/redundancy level; who has access/shared with; permissions and which apps have access/use this data; what is public and what is private!”
2 Likes
Hi Lucas,
Is point 8 still in consideration or are all hosts will have an equal ROI?