I have been doing some research to try and compare the current pricing for cloud computing and do a predictive price comparison for Holofuel.
I don’t know too much about hosting but from what I have researched AWS for example has 160 different price models and different packages alone so it is quite a quagmire of information.
So if anyone can answer my below questions it would be great thanks.
What will be the default or regular period of a hosting agreement between hosts and developers, will the charges (if any) be billed, per second, per use, per 24 hour period or other?
Among the charges for Holofuel is there an Cap on it or is it a set rate for xxx period of time/usage?
Will clients be able to both self host and also pay for hosting at the same time?
How are hosting agreements bound, is it within terms of service or is it a default written agreement created by Holo between hosts and clients?
I have a few more but that is it for now, the main one being the charging methods and also what would be the most comparable hosting service to the AWS pricing model.
Hey!!! i realise you are the famous “WorldOfHolochain” supporter on twitter!! I currently do not hold any port as well. But i am a sizeable investor and support holochain in other ways. I plan to get a holoport during/after beta. I like to see hows the ongoing progress.
Anyway i cant post on holo forum. But i can ask my friends to see if they can post on your behalf.
Hey folks. I’m asking the HoloCentral team what their current plans are for billing. Here are some little bits in the meantime:
There will probably be two separate things that the host bills the hApp publisher for: compute and source chain storage. Both will be similar to on-demand (variable) pricing because you should only have to pay for the compute and storage that you actually use.
Compute is most comparable to Lambda in terms of billing (in terms of architecture it’s different though – with Lambda, a faceless cluster of servers runs a function for you; with Holo host, a host runs a function on behalf of a user). It’ll probably look like microbilling for time increments – for comparison, Lambda bills by exact amount of CPU time, with a minimum 100ms per function call. But payments from hApp publisher to host will be done by lump.
Storage is most comparable to S3 or RDS. I’m guessing it’d be a per-MB sort of thing, billed by the month. The handy thing about source chains is that you only ever add, never subtract, so storage will only ever grow, and in a consistent way. So it could also look like per-MB-added, with a billing cycle starting at the beginning of each source chain write.
What the ‘lump’ period is, I’m not quite sure. I think it’ll look like automated billing; that is, the host accrues charges owed by the publisher until a limit set by the host, at which point they submit a service invoice to the publisher. Maybe Alice has a big custom rig that accrues tons of HF owing per day and wants to be paid out more frequently, whereas Bob has a Nano that is only worth billing for every month.
I believe hApp publishers will be able to self-host, and I believe there’ll be a setup that avoids having to bill themselves (and hence have to pay the transaction fee). Whether they can do both – e.g., self-host the base load (without transaction fees) and let the marketplace host spike loads (with transaction fees) – I don’t know.
Hi @WorldOfHolochain, @Sol, while you are not yet in the HoloForum I’ll bridge the conversation here for you. @pauldaoust already shared quite a bit. I’ll add some other bits and pieces I got from the team. Thanks for the heads up @dcatki!
A couple of days ago @thedavidmeister asked when might we be able to host simple static sites on HoloPorts. @artbrock answered that we are probably not focussing on static asset hosting soon: “HoloCentral veered away from hosting static assets to focus on the APP side hosting. It would be nice to do both, but I’m very on board with the focus being on dynamic apps. It is, after all, one of the main differentiators from all the other crypto storage tools (IPFS, MaidSafe, Storj, Substrate, etc. etc.) and we should play to our strength in that way, or risk the message about our strength getting lost.”,s aid Art.
Following the conversation above. I was sharing my understanding that I thought that early on the Holo differential is hosting distributed apps and bridging holochain apps to regular Internet users. I’m not sure whether we will compete with Amazon at all for quite awhile.
Mary Camacho followed explaining to me that In Azure the service is equivalent to a site hosting. “The static assets are always separate as the ‘blob’ service. Also, if a solution includes many apps those may be packaged together or they may be configured for hosting separately by a publisher. In that sense, our general hApp hosting - could be reflective of many other ‘services’ that are found on cloud providers”, said Mary.
“Imagine having a bot process that runs an automated test on content that is public in the context of an app. Someone might create a bot agent that has API permissions in the DNA of a blog app to review and take certain actions within the app like notify an author, change the status from public to draft, etc. If the blog app were designed such that the authors were opting in to a set of community guidelines, this feature would be a value add… and architecturally, could be a bridged hApp running as a scheduled service. I have no idea yet if that is a best practice in the way that we are promoting architecture of hApps or if that is even the most efficient way – but it is a parallel to how those services on cloud hosting might be reflected in the near term for Holo hosting”, concluded Mary.
