We have no cloud
of our own to sell you.
The constraint first, the provider after.
We resell nothing: Azure and DigitalOcean for machines, Kinsta and Sevalla when a managed platform is what's needed. The choice starts from how you are built, not from a price list.
There is no right provider. There's the right one for your constraints.
This is, simplified, the conversation we have at the start of every project. Tick what applies to you and watch the answer move. Often the answer is stay where you are.
The questions first, the provider's name after.
We have no default answer, and whoever does is selling something. Tick one or two constraints above: the first conversation we have with you is exactly this one, only longer.
What it runs on today, what it costs and what broke last time. Three answers that often decide on their own.
Microsoft Azure.
If the identities are already in Microsoft and there are requirements to document, Azure saves months: the accounts are the ones you have, and it brings the certifications with it. You pay in complexity and on the bill, not in work.
DigitalOcean would cost less and be simpler - but without the identities and the compliance side you're asking for here.
DigitalOcean.
Real machines, a price you can read in one line and no surprise at the end of the month. It's the most frequent choice when you need control over the machine without paying for a hyperscaler's ecosystem.
Azure has more ready-made services, and Kinsta would take the machine out of the picture. Neither is any use to you for what you've ticked.
Kinsta.
For WordPress and the like there's no reason to manage a server: cache, staging, backups and certificates are already in there. Whoever works inside it never has to know what a VM is.
Your own VM would be more flexible, but with these constraints that's flexibility you'd pay for without using.
Sevalla.
Bespoke applications that scale up on their own when traffic grows, with nobody having to watch the machine. It costs a little control and gives back time.
A machine on DigitalOcean costs less, provided somebody keeps an eye on it. Here you said there's nobody.
The bars are a game, not an algorithm: the real choice is made by talking. But the order they move in is the right one.
We build applications in containers, so moving them from one provider to another is days of work, not months. It's why we can afford not to have a favourite.
We don't resell cloud. So we have no favourite to steer you towards.
We keep the machines up. After day one, too.
Set up is the short part. The long one is what comes after: the updates, the backups nobody ever tries restoring, the certificates that expire on a Saturday.
Setting the machines up
It starts from what has to run on them, not from a size on a price list. Network, access, certificates and separate environments are written down somewhere.
Maintenance, not surveillance
Updates, runtime versions, disk space, certificates. Things that make no news until they stop everything on a Friday afternoon.
Backups somebody has tested
A backup never restored isn't a backup. We run the tests, and you know how long it takes to come back up.
Costs you can read
We know what you're paying and why. If a test environment has been running for three months we say so, even though we're not paying the bill.
The first to know something is wrong shouldn't be your client.
On the machines we watch services, space, memory and certificates. On the applications there's Sentry on every project: the error, the line of code and what caused it. The alerts come to us, not to an address of yours somebody has to man.
- On the machines
- Services responding, space, memory, certificates about to expire
- On the applications
- Sentry: the error, the line of code and what caused it
- Who the alerts go to
- Us - not an address of yours somebody has to man
When the site is down an investigation shouldn't have to start.
"That's a server problem, ask them."
"The server is up, it's the code throwing errors."
"Everything shows green on our side."
With software and infrastructure split between two vendors, the first half-day goes on deciding whose fault it is. If we keep both, there's one number and nobody to convince - and if someone else builds the software, we work with them without asking you to change.
If you already have an in-house sysadmin, you don't need us for the infrastructure: we lend a hand when it helps and stop there.
The projects completed over the yearsGo to the case studies