Freelancers & SMBs
Make it work, fast, unhacked.
- a server set up, monitored and kept up to date
- a fast, lean site, no plugins to patch every week
- backups we have actually restored
Linux · Go · Caddy
I make sure your site stays up and your data stays yours.
I write and run Go services that take millions of requests a day on very few machines: fewer servers, less memory, a bill that shrinks. I fix outages, harden your Linux servers, grow your infrastructure with you, and you keep your access and your documentation.
A direct email, no salesperson. Answer within one business day, no obligation.
years of production servers under my care, without a break.
clh --since 1999 production servers, without a break
27 years, no gap.
Run along the trace ⟷
What I sort out for you.
A site that must stay up, data that must not leak, infrastructure that must keep pace. Whatever the size, the same standard.
Startups & scale-ups
Growing too fast, the infra can't keep up.
- a Go service that holds when users pour in, without piling up machines
- automated updates, checked before they go live
- hosting costs you understand and control
Go · Docker · Kubernetes · CI/CD
Large accounts
A specialist on one critical component.
- a delicate migration, seen through to the end
- penetration tests and a remediation plan
- documented work, alongside your teams
DevSecOps · pentest · hardening
We fix it, then hand you the keys.
clh --diagnose your-infra logs, configurations, metrics… ✓ root cause found, explained without jargon clh --plan ✓ clear scope, clear quote clh --deploy ✓ live, backups actually restored clh --handover --docs --training ✓ you own the keys, you don't depend on me
- 01
Understand.
I find the real cause and explain it without jargon: I speak developer, sysadmin and business owner, and I translate.
- 02
Decide.
A clear scope, a clear quote. Free software first: no vendor lock-in when it can be avoided.
- 03
Deliver.
Live, and a backup only counts once it has been restored. Otherwise it's only hope.
- 04
Hand you the keys.
Access, documentation, passwords: all yours, all written down. You don't depend on me.
What I keep running.
01
Take the load for next to nothing
--goMy specialty: Go web services that take millions of requests a day on a modest machine. One compiled program, little memory, no stack of layers: the hosting bill, the power draw and the latency all go down together.
Go · single binary · caching · pprof · efficiency
ProofThis page is served by a Go program I wrote, the same one that runs clh.do and clh.agency.
02
Understand an outage
--diagnoseYour site is slow, goes down or acts strangely: I find the real cause, explain it plainly and tell you what to fix. The problems nobody could solve are my favourites.
logs · configurations · metrics
03
Secure
--hardenI look for the holes before attackers do, fix them and lock down your servers and sites. Security from the first line of configuration, not as an add-on.
audit · pentest · hardening
04
Replace a bloated site
--migrateA site or a store weighed down by plugins, slow, forever needing fixes: I replace it with a tailor-made solution in Go, keeping your content and your page addresses, without taking the service down. One compiled program, faster, far less exposed to security holes, needing much less memory and fewer servers.
content and URLs kept → Go · step-by-step migration
Coming from?How an AI-assisted move works, platform by platform: WordPress, PrestaShop, Magento, Drupal, Shopify, PHP, Node.js…
ProofI designed it for exactly this: NADAPROD, open source, hosted on the same machine as this page. I import your existing site, and every page becomes a self-contained file: no database, no plugins, no trackers, no third-party calls.
05
Build to last
--architectureFrom a single server to many machines, on your premises or in the cloud: a setup that will last ten years, not what's trendy this quarter.
Debian · Docker · Kubernetes · Caddy · Nginx
ProofMy own tools follow the same rule: POSTERM, which I designed, a wall of persistent terminals to run ten coding agents at once, fits in a single Go binary, with no database. Open source (MIT), it runs on the same machine as this page: my coding agents work in it every day.
06
Ship without risk
--ci-cdEvery update goes through automatic checks before it reaches production: fewer mistakes, fewer cold sweats, and security checked at every stage.
CI/CD · GitOps · DevSecOps
07
Host and run day to day
--hostHosting and day-to-day running of your servers (updates, backups, monitoring), on my machines or at datamix.io. Compiled, lean, safe programs: no PHP in production. What I recommend, I run myself.
Go · Python · MySQL · MongoDB
08
Stay one step ahead
--watchI publish a security watch on intelligences.blog: databases left open, software flaws, attacks on AI. What I learn there benefits your servers.
govulncheck · OSINT
OVH? I know the place.
From 1999 to 2003 I ran my own servers, on site. Since 2003 OVH has been my main host: hence the .ovh. Trouble with your OVH hosting, domain, email or server? Answers to the most common errors are here (in French), and I sort out the rest.
ovh trouble
OVH en panne ?
Tout est tombé d'un coup : site, mails ou serveur ne répondent plus.
how to fix it (FR) →ovh trouble
« Site non installé »
La page OVHcloud « Site non installé » s'affiche à la place du site.
how to fix it (FR) →ovh trouble
Erreur 500
« 500 Internal Server Error », ou une page blanche.
how to fix it (FR) →ovh trouble
Certificat SSL
« Votre connexion n'est pas privée », ou le certificat ne se génère pas.
how to fix it (FR) →ovh trouble
DNS et nom de domaine
Le domaine mène nulle part, ou vers l'ancien site.
how to fix it (FR) →ovh trouble
Mails en spam
Vos mails arrivent en spam, ou sont rejetés par Gmail.
how to fix it (FR) →- all OVH troubles (FR) →
- independent · not affiliated with OVHcloud
Machines change. The service never does.
Since 1999 there have always been production servers under my responsibility: those of the companies where I was DevSecOps and often the only sysadmin, my clients' and my own. My own on-site servers from 1999 to 2003, then OVH since 2003. Linux at home since 1996; I remember when everyone ran their own DNS resolver.
1996
bedrockFirst Linux boxes at home, first website. A year later, first paid web service.
1999→2003
AlynetCoder, then lead coder, then CTO: the NoSQuiz and AlyMail ad networks, and RedirectWeb. A platform competing with DoubleClick in Europe, until Google came in.
l3dlp.com, registered 16 June 2000, still online2003→
OVHNo more on-site servers: OVH becomes my main host, and still is. Shared hosting, VPS, dedicated servers, Public Cloud, domains and email, for me and my clients. Hence the .ovh.
OVH troubles, and how to fix them (FR) →2003→2010
MarseilleAt the Friche Belle de Mai, within the Sparrow studio, and for Radio Grenouille: sites, platforms and hosting for institutions, festivals, theatres and companies.
Région PACA · Friche Belle de Mai · Radio Grenouille · MailInBlack · Maxicours2010→2014
MediaStanzaDevSecOps and the company's sole sysadmin.
2015→
datamix.io · card.pmCofounder and director of datamix.io, a privacy-first host. And the servers behind card.pm.
2019→
clh.ovhThe infrastructure and security practice, direct.
2026
todaySome fifty sites and services behind the same server as this page. Updates, backups, logs and security: hand-kept, every day.
A problem,
a project?
Describe the situation, even if it's only a doubt or a precaution: I answer within one business day, no obligation.
Describe my problem→or write directly to cedric@clh.ovh