They understand that if anything goes wrong I will offer a complete refund (which is $0), because they’re my friends. I couldn’t imagine doing it for other people though.
It also helps that they’re IT or at least interested in IT.
It's free and comes with no guarantees. If you need a domain or subdomain, you own it, pay for it, and just put a DNS record to the IP I tell you.
Usually you only need some subset of the data per page load if you invest some time looking at dev tools you can probably find the API call you need and save yourself a few MB.
They all offer scraping APIs now too that can be cheaper for certain use cases where you only need a subset of the data that is actually loaded. Like $1.3 per 1k requests: https://oxylabs.io/products/scraper-api/web/pricing or $1 per 1k: https://brightdata.com/pricing/web-scraper
https://faq.oxylabs.info/en/articles/8826164-restricted-targ...
Additionally I believe that $4/GB is an introductory price. When I went into the Oxylab dashboard, it showed me $8.
I’m curious if you can suggest a happy medium between curl-impersonate on VPS (dirt cheap) vs residential proxy ($8/gb)?
Personally I’m not trying to hit any of those common sites.
Most of them seem pretty reasonable?
"Entertainment & streaming" - who's trying to scrape netflix's library?
"Banking and other financial institutions" / "Government websites" / "Mailing" - seems far more likely it'll be used for credential stuffing than for "scraping".
"Ticketing" - seems far more likely that it'll get used by scalpers than for scraping
The main targets of scraping - e-commerce sites (for price comparisons) and social media networks (for user generated content) are fine to scrape. Is there some use case I'm missing here? Is there a huge contingent of people wanting to scrape ticketmaster or bank of america?
https://www.courthousenews.com/wp-content/uploads/2024/01/me...
"Residential proxies is based on traffic and purchase model. Pay as you go model starts at $7.35 per GB, and can be discounted as low as $1.84 per GB when purchased in bulk."
But I would guess this type of proxies is mostly use to send data rather than receive. You can access geo-locked sites through standard vpn.
This is more than I expected a hobbyist to be able to pull off. How/where would you market to achieve this cash flow? If I could make $$$/mo with 1 computer at home, could I scale to $$$$/mo by adding compute/storage?
The target is $100/m per RTX 3090.
Self-hosting a service, website, or SaaS happens alot more than most people realize especially with fibre to the home, and even things like Starlink.
Sharing your home services doesn't seem mission critical, unless it is. It's easy to just pop in a QNAP or Synology for any individual for themselves.
Businesses who have data residency and processing requirements are live and well, and pay a premium.
The pendulum of cloud vs local is starting to swing the more AI models become available in datacentres to keep close to your data in 2025. Google just announced privately hosting to Gemini a few days ago iirc.
For self-hosting, one has to stabilize power, internet, and decide on equipment. This is orders of magnitude more doable, easier and cheaper every 5 years going back 15.
Just have to be clear on how experienced you are with this in the past and most recently. The cloud is ridiculously profitable and overpriced because in part of how much this has come down market just not used.
Of course there's knowledge involved, and again, there's plenty of people who quietly have this knowledge, and it's far easier to obtain now starting with something as simple as Proxmox and/or Docker.
There’s collocation for mid tier usage patterns that’s cost effective maybe but I imagine it’s on the decline in general these days, and it never seemed that cost effective for hobby stuff unless you had a group of people who were splitting it.
Can you expand on this? I'm in the US and have 1 Gbps symmetrical fiber for $70/month. The US is a big country and some people will have it very slow, but usually not the same type of people who self-host.
This is a typical offer in fiber now in France.
Not that it matters because my equipment is 1 Gbps but at least I can flex :)
1. enthusiastic enough to have their own homelab and host their own kit.
2. have money and willing to spend on 'proper' hosting - either AWS/Azure/Hetzner/OVH style or cheap VPS style. You say "few hundred bucks a month" but a low quality virtual server starts from from $1/month - https://lowendbox.com/blog/1-vps-1-usd-vps-per-month/ - and will be in a datacenter. You'd need a lot of customers or a very competitive offering to make even one hundred bucks profit every month.
3. have no money or don't want to spend money, using Oracle/Amazon free tier, SDF free shell, a free account from a friend with a homelab, etc. They don't make good customers.
That leaves people who have money and are willing to spend it on a low quality product instead of a higher quality product, which is basically your third option - friends who are giving you money and putting up with a worse deal out of friendship - charity, basically.
To wit, no serious business would use your services, so your market is pretty much limited to:
1.) Amateur hobbyists (very little money there)
2.) People with Bad Intentions (horrendous from a legal and security perspective, and ethics if that’s your thing)
3.) People in your social circle