Technote

Servers & infrastructure Intro

Series Web infrastructure on a shoestring Part 2 of 8

Shared hosting to a VPS: when is it actually time?

You do not move because a VPS is better — you move because something is blocked. The signals are specific, and what grows after the move is responsibility, not just the bill.

Shared hosting is an excellent starting point. It is cheap, it asks nothing of you technically, and it is genuinely enough for most early sites. The issue is timing, not the tool — as a site grows, there comes a point where the shared environment is not slow so much as blocking.

“It feels a bit slow lately” is a weak reason to migrate, though. Slowness far more often comes from images or plugins than from the server, and moving in that state simply gives you the same slowness somewhere more expensive. Judge on different signals.

The signals that say it is time

None of the following are problems you can pay your way out of. They are things a shared environment structurally cannot do. When two or three apply at once, it is time to look at moving.

A shortage of permissions, not of budget — money stops postponing it here

Add one more: getting slower because of a neighbour. Shared hosting puts many sites on one machine, so response times can wobble while nothing on your side has changed at all. The cause is not in your site, nothing you fix reproduces it, and a repeating pattern of that is itself the signal.

What grows instead of the bill

A VPS is permission you buy, and permission arrives attached to responsibility. Everything the host quietly did for you becomes yours. Leave that out of the calculation and you save on the rate while spending far more in hours.

Four things in particular become yours. Operating system security updates now need applying by you; firewall and access control need configuring, because the defaults are not safe; backups do not run on their own; and failure detection starts with simply noticing that something has stopped. The hours those four take usually exceed the difference in price.

Which leaves a founder three realistic options: stay on shared hosting and attack the bottleneck another way; move and learn the four items above; or move but have the build and monitoring handled. Which one is right is decided by the value of your hours — unless learning is the point, work out whether the time to learn costs less than the price of delegating it.

If you are moving, the order matters

A migration does not finish in a day. Keep the order and the situation stays reversible; skip a step and you discover the problem only after the address already points at the new machine.

Four steps — leave the old server running for a few days

Respect that last box especially. DNS changes do not propagate instantly, so for a few days visitors arrive at both servers. Delete the old one during that window and every enquiry that landed there is simply gone.

If you have not yet mapped the line items, start with what hosting really costs; broader server material lives in the Servers & infrastructure archive. Handing the migration and the optimisation over as a single job is what our optimization program is for.

Next part

With the infrastructure decided, the tools that sit on top come next. The following part is about telling free-is-enough from worth-paying-for — judged by the weight of your data, not by the feature list.

More on this topic

All technotes

Servers & infrastructure Practical

Sending email yourself, or handing it over

Run your own mail server and the invoice is near zero — you pay in deliverability instead. Which is cheaper depends on what one enquiry is worth to…

Founders 7 min read

Servers & infrastructure Practical

Knowing your site is down before your customers do

If a customer tells you the site is down, you are already late. Deciding what to watch and how to be alerted takes about half an hour.

Founders 6 min read

₩270,000 · Join the program