I write a lot about self-hosting on this blog. Linux home servers, Immich for photos, n8n for automation, Proxmox, the works. It’s a whole cluster of posts, and readers seem to like them.
Here’s the part I haven’t said out loud until now: I don’t self host everything I write about. My own production stuff runs on someone else’s servers on purpose.
Not Immich. Not Nextcloud. Not a Jellyfin box in my closet. My family photos live in the same cloud service yours probably do. My blog runs on hosting I pay a company to babysit. My SaaS products sit on a managed platform that deploys on git push. That’s my actual stack, and I’m not embarrassed about any of it.
And that’s not a contradiction. Let me explain why.
What I actually run
Here’s the real stack, no marketing spin.
bigguyonstuff.com (this blog): Running on a DigitalOcean VM. That’s it. Nothing else in the path. WordPress sits on the droplet, I don’t babysit the server, I don’t patch kernel CVEs at 11pm. When something breaks, I have a handful of known fixes or I open a ticket.
pickmysquare.com and other side-project SaaS: Render.com. Git-push deploys, managed Postgres, preview environments, free tier for the low-traffic pieces. I write code, commit, push, and it’s live.
Trigli (my AI customer support SaaS): Also on Render. It runs a real customer workload with real uptime expectations, and Render handles the infrastructure I don’t want to think about at 2am.
Experimentation sandbox: I have a Hostinger account I use for messing with stuff like n8n and OpenWebUI. That is not production. It is not in the path for this blog or any SaaS product. It’s where I break things on purpose.
Notes, photos, docs, email, calendar: Google, Apple, and the usual suspects. Not self-hosted. Yes, I’ve seen the Immich and Nextcloud posts on this blog. I wrote those as beginner guides, not as what I’m actually running. More on that in a minute.
Home lab: I don’t have one in production. I might build one someday for learning. I don’t run my business on one.
That’s the whole truth. No secret Proxmox cluster in a closet. No Pi running TrueNAS for the family photos. Production is DigitalOcean plus Render, sandbox is Hostinger, and that’s all. Just managed stuff, running quietly, while I work on the things that actually make me money.
The real math, not the blog-post math
Self-hosting tutorials love to do the math that ends with “and you save $40/month!” The math usually stops there. Mine doesn’t.
Here’s what a self-hosted production stack actually costs when I do the full tally:
The hardware or VPS: Yes, this is cheap. A used mini PC or a $10/month VPS gets you the OS. That’s the part every tutorial covers.
The time. Patching. Monitoring. Restoring a backup when the SSD dies. Figuring out why the reverse proxy stopped renewing certs. Running a serious self-hosted stack is a handful of weekend hours per month, and those hours never land at convenient times. Saturday morning. The night before a Trigli customer demo. During my kid’s baseball game. Enough hours per month that it starts to feel like a second job I’m not getting paid for.
The security debt. Security researchers have documented tens of thousands of internet-exposed self-hosted instances with unpatched vulnerabilities in the months following major CVE disclosures, including the February 2026 wave. Those breaches don’t happen because the people running them are dumb. They happen because they got busy and a security update slipped past a weekend. I would be one of those people.
The “I don’t want to think about this” tax. This one doesn’t fit on a spreadsheet, but it’s real. I run a SaaS company and a blog and I have a family. My brain has a budget. Every service I have to maintain spends some of that budget. Managed hosting buys me the right to not-think-about things. At my hourly rate for my actual work, $15/month to not-think-about my blog is cheap.
Now imagine I had 10 self-hosted services. Multiply the maintenance. Multiply the outage risk. Multiply the weekend phone alerts. That’s a part-time job I’m not getting paid for.
Where managed is obviously the right call
If you’re in any of these situations, managed wins on basically every axis that matters:
- You ship to customers with any kind of uptime expectation. If somebody’s paying you and expecting the service to be up, you don’t want to also be the sysadmin who was asleep when it wasn’t.
- The managed tier is free or nearly free. Render’s free tier handles a surprising amount of traffic. DigitalOcean’s entry tier starts at $4/mo, with $6-$12 as the realistic floor for a small WordPress workload. When managed costs less than the power bill on the mini PC, the self-host math falls apart.
- You’d rather sleep through a kernel CVE. This is half-joke, half-serious. I do not want to wake up to a security alert and spend my Sunday patching an exposed service.
- You don’t want to be the self-hoster whose home server quietly got compromised three months ago. Managed providers have security teams. You do not. Be honest about the skill gap.
- Your business income is worth more than the hosting savings. If you can’t cover the managed cost from the revenue the project makes (or hopes to make), the project might not need a server at all.
Where self-hosting is still the right call
I’m not anti-homelab. Some situations self-hosting genuinely wins, and I try to be honest about that in the posts I write about it.
- Learning and skill-building. This is the biggest one, and it’s why the blog’s self-hosting posts exist. Standing up Proxmox, configuring Docker Compose, wiring up Traefik, fighting with certificates. All of that is real engineering practice. If you want to get better at Linux and infrastructure, hosting real services (even if they’re just for you) is the fastest path.
- Privacy-sensitive stuff at scale. If you’ve got a decade of family photos and your alternative is paying Google or Apple $10/month forever, an Immich box starts looking reasonable. Same with notes, documents, and other stuff you’d rather not leave with a third party.
- Hobby projects with no revenue. If nobody’s paying you and nobody’s counting on it, who cares if it goes down for a day? Self-host it. Break it. Fix it. Learn.
- Workloads with weird requirements. Sometimes managed platforms won’t let you do the thing you need (GPU access, unusual network setups, custom kernel modules). Self-hosting is the escape hatch.
The pattern: if it’s fun, educational, or your own personal data with no stakes attached, self-host it. If a customer, employer, or your own livelihood is counting on it, pay somebody to babysit it.
The hybrid I actually use
Here’s how the two worlds coexist in my setup:
Production lives on managed infrastructure. The blog, the SaaS apps, the customer workloads. That’s DigitalOcean for bigguyonstuff.com and Render for Trigli, pickmysquare, and the other SaaS stuff. I’m not moving any of them to a homelab anytime soon.
Experiments live on a sandbox account or disposable cloud VMs. Hostinger is where I mess with self-hosted tools like n8n and OpenWebUI without touching production. If I want something more disposable than that (a weird framework, a test database, a Claude Code setup for a client project), I spin up a cheap DigitalOcean droplet, mess with it, and destroy it when I’m done. Total cost: usually under a dollar. No physical hardware, no permanent maintenance, no guilt when I delete it.
Learning projects could live on a future homelab. If I build one someday, it’ll be for learning and personal data, not for production SaaS. That separation matters. Mixing production and experimental on the same hardware is how you get outages that also take your kids’ photos offline.
That split is why the contradiction isn’t real. The homelab posts on this blog are for a reader who wants to learn, experiment, or self-host personal-use data. They’re not claiming I run a homelab in production. I don’t, and I won’t pretend otherwise.
Why the homelab cluster still exists on the blog
Fair question if you’ve read this far: why publish all those self-hosting posts if I don’t run any of it?
Three reasons:
- Readers ask for it. Self-hosting is genuinely popular. Plenty of this blog’s audience wants to learn Immich, n8n, Proxmox, and I’d rather help them do it well than ignore the topic.
- Writing a beginner guide is itself a reason to stand up a lab. I’ve set up Immich to test the post. I’ve run n8n in Docker. I’ve configured Home Assistant dashboards. Those were tutorial-test setups, not production migrations. And I’ve been clear in the posts that they’re beginner guides.
- The tradeoff deserves both sides. Most tech blogs push one narrative: self-host everything, it’s amazing. This post is the counterpoint. Both can be true. Which side fits depends on the reader’s workload, not a tribal allegiance.
If you came here from one of the homelab posts wondering if I secretly run a rack of Dell PowerEdges, I do not. I run a DigitalOcean droplet and a Render account and I sleep fine.
What about cost?
I’m not going to quote my exact bill because pricing moves and this post should age. But at public tier pricing in 2026:
- DigitalOcean: Droplets start at $4/month for the entry tier (1 vCPU, 512MB RAM). Realistic floor for a WordPress workload is the $6-$12 range, and a site with real traffic lands around $12-$24/month. DigitalOcean also switched to per-second billing in early 2026, which benefits workloads that don’t run around the clock.
- Render: Free tier for low-traffic web services. Starter paid tier around $7/month per service. Managed Postgres adds more, scaled by the size.
- Hostinger (my sandbox): Shared plans run in the $3-$10/month range, VPS higher. I use it for experiments, not production, so this line item is optional for most people.
Add it up for a solo founder running one blog and a couple of SaaS apps and you’re in the ballpark of $50-$150/month total, depending on scale. That’s less than the hourly cost of one day of my time to debug a broken home server. And it’s a predictable line item I can forecast against revenue.
One independent 2026 writeup that tracked a solo self-host stack found an average cost of around $71/month (hardware amortized, power, bandwidth, and sundries). Once the author priced their own maintenance hours at even $50/hr, the “cheap” self-host option ballooned to $250-$400/month in real cost. Equivalent managed setup: about $49/month flat. Their numbers match mine, and I don’t even price my own hourly at $50.
Quick note: cloud hosting pricing has been shifting in 2025-2026. DigitalOcean and Render have both moved their entry-tier floors. Whatever numbers I quote today will drift. Check each vendor’s pricing page before committing.
What I’d tell a solo founder or side-project dev
If you’re running a business or trying to ship something real, use managed hosting. Give yourself the gift of not being your own sysadmin. Buy back the brain cycles and spend them on the actual thing you’re building.
If you want to learn Linux, infrastructure, networking, and system administration, build a homelab. Run it for your own data, for learning projects, for fun. Do not conflate “I self-host my photos for privacy reasons” with “I should self-host my startup’s production database.” Those are different decisions with different risk profiles.
And read both sides of the topic before committing. The homelab posts here on the blog are legitimately useful if self-hosting is where you want to go. This post is the other side of the aisle. Pick the path that matches your actual workload and stress tolerance.
What’s your real stack?
Drop a comment with what you’re actually running. Not the aspirational setup, the real one. I’d especially like to hear from anyone who went full homelab and regretted it, or went full managed and wishes they’d self-hosted more. Both directions are useful context for the next person reading.
Share this with a friend who’s been wavering between a $500 mini PC and a Render account. And if you want a walkthrough of how this blog’s content pipeline actually runs on Claude Code plus managed infrastructure, I already wrote that one.
Sources
- Qovery – Self-Hosted vs Fully Managed Hosting for Mid-Size Teams
- UpCloud – Self-Hosted vs Managed Databases 2026 Guide
- Pantheon – Managed vs Self-Hosting Learning Center
- Liquid Web – Self-Hosting vs Managed Hosting for MSPs
- Render pricing
- DigitalOcean droplet pricing
- Hostinger plans
- dev.to – Managed Hosting vs Self-Hosted in 2026: The Math Surprised Me
Pingback: Linux Home Server Under $200: Honest Beginner Build (2026)