Building and operating a public cloud from Kigali
Friday
22 February 2019
Official launch — Rwanda Coding Academy
“Your website is dry.”
Someone walks up. Offers feedback. Changes everything.
Awesomity = awesome creativity.
So we became Stunnity.
Stunning creativity. We were not subtle people.
- stunevents — failed
- stunlab — failed
- stunpark — failed, and the idea got copied
- RCA website v2 — failed
- Kurious, e-learning — failedbuilt with Cedric, who is in this room and is our founding engineer today
- Client websites — evolve.rw and others. What kept us alive.
- Signa — QA
- MINADEF · NISS — e-learning. Full-stack, then team lead
- BP — senior frontend engineer
- Pivot Access — traffic fines, driving examsif you've ever paid a traffic fine in this country — you're welcome
- RSSB — imisanzu and ishema
- Strettch → Strettch Cloud
We knew a lot.
And we were bored.
No capital, no idea. So we started an agency to get both.
It didn't feel like an idea.
It felt like a diagnosis.
So we pivoted. Started SC. Quit our jobs. All in.
You have exactly two hosting options.
Both make you compromise.
The global clouds
Local providers
The global clouds
- Your users are here. The servers are not.every request commutes to Europe and back
- You pay in dollarsyour CFO hates it, your margins hate it more
- The law increasingly wants certain data stored in Rwandaso now compliance is your problem
Local providers
- Servers are here. Everything else is missing.expensive, slow, limited services
- The operating model is a human beingyou email someone. you WhatsApp someone. you call someone to reboot a server.
- No transparency when it breaksyou find out your VM is down from your customers
A cloud platform without the compromise.
Infrastructure as a service, built and operated from Kigali.
- Self-service. You operate your own cloud.no emails, no WhatsApp. about two minutes.
- Servers in Rwanda.compliance is a checkbox, not a project
- RWF, hourly.destroy a VM after an afternoon — pay for an afternoon
- More services than any local provider
- A team that tells you the truth when things break.
October 2025
We launched. Private beta.
266,099RWF
Our entire first month of revenue.
5,317,057RWF
Last month. 20×, nine months later.
cloud.strettch.com
Four things you never configured
Its own VLAN
Every team gets one. Your network never touches another customer's.
Metadata service
Hands each VM its identity at boot. AWS-style.
The agent
Ships metrics before you've even SSH'd in.
The meter
Honest to the hour, from the second it boots.
AMD EPYC · 2TB RAM each
SONiC switches
EVPN-VXLAN
per server
+ Ceph EC tier for backups
with an N+1 host in reserve
Team VMs deliberately spread across hosts — so one dead server never takes out one customer's whole product.
“A datacenter cannot lose power.”
Two grid feeds. Generators. Batteries. UPS. Five layers. On paper it can't.
- 00:05Both power feeds fail at once. No warning from the provider.
- 00:45Power is back. The outage lasted forty minutes.
- —Control plane reboots believing every VM is still “Running”. It restarts nothing.
- —Alerting can't scream. It's dead too. Same building, same blackout.
- —A VPN route hijacks the control plane's routing on the way up. It can't even talk to the hypervisors.
- —We are all asleep.
- 06:00Found on a routine check. Two hours to untangle it.
40 min
The datacenter lost power for forty minutes.
7 hrs
We were down for seven hours.
That difference — that's design. Monitoring now lives in an independent zone. Someone else's building.
Our customers don't host websites on us.
They host their customers on us.
Our downtime is their downtime, squared.
The December outage
Every affected customer got 24 hours credited, automatically.
We published the full postmortem — including the embarrassing parts you just heard.
June · MTN forwards us two reports
We scanned our entire public address space — including the block nobody monitors on our behalf.
Exposed databases. Leaked credentials. Worse. On customers' machines.
43 notices, one day
Then a day as a 2-gigabit DDoS cannon pointed at China.
RAM dies. CVEs drop.
Bots never sleep.
Every day this field hands us a problem we did not predict a month earlier. And honestly? That's why we love it.
What we don't have
- No SLA yet. Compute uptime last quarter was 99.8% — but we won't sell you a promise until we can back it with automated failover. Today you get the number, not the contract.
- No managed services. No managed databases. No Kubernetes. No object storage. No one-click backups.
- AWS has 500+ services. We have two that matter. We know exactly how short that list is.
- Uptime guarantee — a real SLA
- Dual-stack IPv6 on every VM
- The 50-gig east-west fabric you just saw in Gen 1
- Managed Kubernetes
- Storage as a product — volumes and S3-compatible object storage
“It's dry.
Give it life.”
You don't get to choose where a spark lands. This one landed on three students.
Your clients' apps belong in Kigali.
Not your client in Europe — I won't pretend otherwise. But anything government-adjacent, fintech, health. Anything where the data must stay in Rwanda, or where users feel every millisecond of the round-trip to Frankfurt.
cloud.strettch.com
Clarance Liberi Strettch Cloud [email / phone]
One SC team holds all your client workloads, with your engineers as collaborators.
First project's credits are on us.
[decide the real number]
Someone here told us to make it wet.
We made it rain.
Thank you
cloud.strettch.com · [email / phone]
Questions
Down-arrow for the five you'll probably get.
“Why not just resell AWS locally?”
Data residency, latency and RWF billing aren't resellable. The compromise is the gap.
“Whose datacenter is it?”
Colocation for power and space; everything above the concrete is ours. The December 16th story already shows we don't hide behind them.
“How do you compete with Hetzner's prices?”
We don't. Hetzner has no servers in Kigali. We price latency and compliance, not scale.
“Most of our work is for remote clients.”
Agree, don't argue. Those aren't the projects I'm asking for. Then the wedge: dev and staging on SC — hourly, RWF, disposable. Fits every project regardless of where production lives.
“What happens to my data if SC dies?”
⚠ Write the real answer before stage day. Hardest fair question in the room. “We're committed” is not an answer — exports, escrow, standard image formats, a documented exit path.