Would eventually the goal of IPFS be to make cloud services obsolete and provide each human being with home bare metal servers at home to share and receive content from? That is, people can host their content via their own home server and websites are hosted by their own servers. Or will this be too expensive, too technical, not feasible and the big centralised hosting services that provide storage and bandwith to central content sharing platforms will stay in the end ?
Raspberry PI + SSD = 24/7 IPFS node for cheap. That’s what I use and it could be made really easy for normal people.
On the other side, some people will pay for convenience.
Oh please no. I completely understand the desire to do so and there is nothing stopping people from doing that just like there’s nothing stopping people from self hosting. (I did that myself for a long time) but I really don’t think that should be a universal goal. After having done it for a while I have come to the conclusion that your home is not a datacenter. I live outside two major metropolitan areas and the electricity flickers when there’s a storm. I have my stuff on an UPS but the batteries have to be replaced periodically and they’re not particularly cheap. Home AC systems weren’t setup for handling the kind of heat generated by high end servers and neither is the wiring. Alternatively you are stuck with low power systems like the Pi.
Now you’ve got the security nightmare that is the home server. Someone has to maintain it and keep it up to date. In the end you just end up with someone else’s computer in your basement and you’ve got the same situation before but worse because, your house is not a data center.
I feel like focusing on running something out of your home misses the mark just like people used to say, “everyone should learn html and write their own homepage”. No they didn’t. So they handed everything over to Wordpress, Blogger, Google, Facebook, etc. I wonder what would have happened if someone had said, “People don’t want to write html. They want social media and to be able to communicate and express themselves”.
Thank you for your answer @zacharywhitley. I understand your point of view and see the problems and hassle it gives to host your own server at home. However, I just find it difficult to see how IPFS can shine if people need to host their own 4k videos for example. There are pinning services like pinata, infura and others that offer to do this for you against a price. However, if they would get really big then how decentralised and distributed is the content if it is not cached frequently enough by users and if the laptops of users that did pin it are turned off and thus cannot provide the request but only the pinning service. According to me, but correct me if I am wrong a pinning service is basically the same as a centralised database, the only difference is that it allows incoming IPFS requests.
Do you think that IPFS + pinning services is then the future for the future content creators and or platfroms ? and only a small amount of people will actually have 24/7 machines running at their home?
How many people are running their own machine 24/7 now? I know being in IT makes it seem like a lot but I’m willing to bet that if you took an objective look it’s an almost insignificant number.
Need to get away from the paying $$$ == bad. Look at Facebook. They’re frickin’ evil and it’s completely free to use and that’s probably one of the problems. You pay for all kinds of things you use. This shouldn’t be any different. I think focusing on free to the exclusion of everything else is not good. Even if you self host it’s not free. You’re paying for the internet connection, the computer, the electricity, cooling. I think IPFS should be shooting for cheap and ubiquitous not free. Free should just be one of the options but you’re going to find yourself paying for free but probably in some opaque way.
Maybe one way you can look at IPFS would be like hosting and bandwidth insurance. Insurance costs a premium above expected losses but what you get for that premium is consistency. You pay the expected losses not the actual losses which have really high variability.
When you host something with IPFS you’re paying the expected costs of hosting and the big spikes get smoothed out. Most of the stuff I’m going to post isn’t going to get any traffic but if I get a spike I’m going to be hammered. If you look at going viral like a car accident you spread out the costs of going viral among everyone and in return everyone pays a small price and has consistently available content and hosting.
Thanks again for your reply. I understand that financially speaking hosting your own content is not for free and not feasible at the moment. However, I wanted to look at it from a technical point of view since eventually the price of electricity, disk space, bandwith etcetera will all go down due to science and engineering. So if it becomes really cheap to have your own server at home and it will be just as easy as to install a tv in you house will it then be the best infrastructure to have for sharing information on earth ? Also one could share files with friends as backups on different locations. Will pinning services and regular clouds still be necessary even in this case ?
I run several services hosted on home based servers running 24/7 … I typically forward those services to an Internet facing IaaS server. A home based IPFS server should be fine if some of the high bandwidth/high availability material is also pinned using a “cloud” based service.
What would be most useful for my own purposes is a method to distribute IPFS data across a private IPFS network in a controllable predictable way…
- Home_server_1 has 10% of the network data
- IaaS_server has 40% of the network data
- Home_server_2 has 50% of the network data
If and/or when such controllable distribution can be set, self hosting content could be extremely inexpensive by placing large but unpopular files on home based SSDs that may go offline every once in awhile.
RPI max memory is still 8 GB? With some light OS, you can have max 2 nodes on it.
PIs are so cheap just buy more!
Why more than one node though? Does it not scale well?
IPFS DHT do not scale well. Large nodes do not have enough time to announce all their blocks to network. You need to announce each block at least once per day otherwise other computers won’t find the content you are sharing.
In 0.9 there is alternative DHT code which addresses scalability and it can be turned on for testing purposes.
Hi! You may be interested in A preliminary discussion of the “Workswap” protocol . There I’m proposing the idea of a self-governed network with peers providing services to each other. So, it’s about self-hosted services too. Thanks in advance for any comments and suggestions!
I think that would be great. If people could easily and cheaply pin content they find on the internet on their own computer, it could become a habit and culture.
The thing to do is make it “one click”, so easy that it is as effortless as using Big Data solutions.