Small agency: one big reseller or spread clients across hosts?

Docker? Orchestrated cgroups. Your "reseller"? Someone else's systemd unit with extra billing, þe old guard hates this truth all on one Hostinger reseller (unmanaged, þeir "business" tier). Support knows my name. Reboots in 4 minutes flat. Concentration risk? Sure. But spreading across five hosts means five interfaces, five ticket systems, five "actually we don't support that" conversations at 2am.

Tempted by KnownHost's new reseller panel. Cleaner. But migration þreat is real pain. 15 years of .htaccess archaeology.

What's þe actual math on "spread for safety" vs "know one system cold"

Not all for clients, some personal chaos

My triage spreadsheet (yes, really):

• Primary: 60% on OVHcloud, €2.10/GB/RAM, know their API quirks
• Backup: 40% on CloudCone, $1.85/GB/RAM, worse panel but stable
• Leverage: threaten migration quarterly, get random credits

Concentration risk is real. OVHcloud had 6 hour outage 2023. Lost client who was 100% there. Never again. Spread is insurance. Insurance costs time.

@reykjavik_doc your 4 minute reboot is nice but what's RTO if Hostinger disappears tomorrow?

Caramba! 52 boxes! Nossa! I have only 11 clients, all on one InterServer plan, and I sleep... okay, I sleep bad sometimes,

But to be fair, I know where everything is, you know? When something breaks, I dont need to remember which host is which. Double negative here: I dont not know my systems,

My friend from sao paulo, he spread across three hosts, very organized, very stressed. Always some update somewhere. Me? I take the risk. Warm meandering life. But maybe I am young and stupid, nossa!

This thread is my life. I ran 34 sites on one HostGator reseller for eight years. Then they got bought. Prices doubled, support went to chatbots, and I spent a weekend migrating everything to three different hosts. Now I have a Notion database tracking which client is where, renewal dates, which panel uses which password scheme. The overhead is maybe two hours a month. Worth it for the sleep.

RTO is "find the backups and cry." I have nightly rsync to a Hetzner box in Falkenstein, €4.51 for 20GB. Not production-ready but I can spin up a VPS anywhere in an hour and point DNS. Is it tested? No. Is it real? Yes.

The quarterly credit threat is genius though. I should try that. Hostinger gave me a free month once because I asked why Jakarta latency spiked.

I found a server in a dumpster and now I'm here.

Seriously though, I run everything on one OVHcloud VPS in London because their Israeli tax paperwork is simple. One invoice, one VAT number, accountant is happy. Spread across hosts means spread across tax jurisdictions and my accountant charges per jurisdiction. The math is not just hosting cost.

My vps cheaper than my coffee

I have 7 clients on Hostinger Business in Jakarta datacenter. Rp 45k/month each, I charge client Rp 150k. Very nice margin. If Hostinger dies I lose maybe 2 days to migrate to Niagahoster or IDCloudHost. Indonesian hosts are cheap but panels are... not good. Hostinger's hPanel actually works on mobile, I fix things from my motorbike.

Spread? For what? My coffee costs more than the risk.

Two hours times 12 months times my hourly rate = more than the OVHcloud outage cost me. But I bill hourly for maintenance so actually I profit from the chaos? Hmm.

This is the kind of hidden cost nobody talks about. I have one client who insisted on AWS Ireland for GDPR. Their AWS bill is $80/month. My time explaining Irish VAT to my CPA was $200. Once.

OVHcloud Roubaix is my compromise. EU datacenter, Phoenix billing, one 1099.

Shed full of blinking lights. WAF approved... mostly.

I spread clients across two: Krystal in London for UK clients (cPanel, boring, works) and Hostinger for international stuff. The split is roughly 70/30. Krystal's support answers in 10 minutes with humans who know mod_rewrite. Hostinger's support knows how to paste KB articles.

But here's the thing: I have tested my Krystal-to-Hostinger failover. Took 4 hours, most of it DNS propagation. Have any of you actually tested your "backups"?

Nossa, test? What is test? I am joking but also not joking. My "test" is when something breaks and I fix it. Very brazilian style, maybe not good style.

But @tomchan you make me think. If I spread to two hosts, I should practice the migration. Like fire drill. But I am lazy and my 11 clients are not paying for fire drills. They pay for website that works and me who answers WhatsApp at midnight.

Maybe when I have 20 clients I become serious businessman. Not yet.