Server Sales
Server Sales
Dedicated servers and VPS sized for the workload you actually have — specified with you, delivered configured.
Sized for the job, not for a pricing page
Most server purchases go wrong in one of two directions: far more machine than the workload needs, or a machine that fits today and nothing after it. Both come from starting at the price list instead of the workload.
We work out what you are actually running first — traffic pattern, database size, storage growth, backup window — and recommend from there. If the honest answer is a smaller machine than you expected, that is the answer you get.
How we scope it
- Workload. What runs on it, how many concurrent users, what the traffic looks like at its peak rather than its average.
- Data. Database size today, growth over the last year, how much of it is hot and how much is archive.
- Constraints. Where the data is allowed to live, what has to be isolated, what the backup window can be without affecting users.
- Recommendation. Dedicated hardware or a virtual server, with the reasoning written down rather than asserted.
- Specification in writing before anything is ordered.
Dedicated or virtual: the real difference
A dedicated server is single-tenant. The chassis, the CPU cores and the memory bus are yours, with no virtualization layer between your workload and the hardware. A VPS divides one physical machine into several isolated instances; the resources are allocated to you, but the host underneath is shared.
| VPS | Dedicated server | |
|---|---|---|
| Tenancy | Shared host, isolated instance | Single tenant |
| Noisy neighbours | Possible on heavy I/O | Not applicable |
| Virtualization overhead | Present | None |
| Scaling up | Often a resize | New hardware |
| Cost | A fraction of dedicated | Higher, predictable |
| Best for | Most web workloads | Sustained I/O, licensing or isolation requirements |
For most business websites a VPS is the correct answer and the difference is invisible. The case for dedicated hardware is usually one of three things: sustained disk I/O that a shared host cannot guarantee, software licensed per physical core, or a requirement that the machine not be shared at all.
Where server budgets actually go
The advertised monthly price is rarely the final number, and the gap is predictable. Control panel licences, backup storage, additional dedicated IP addresses and premium support tiers are the usual additions. Server location matters too — physical distance adds latency, so a machine near your users beats a marginally faster one far away.
We quote the whole line, including the parts that show up on the second invoice rather than the first.
Delivered running, not as a bare box
The server arrives configured: operating system installed, firewall closed down to what you actually need open, users and access set up, backups scheduled, monitoring connected. You get the documentation of what was done and the credentials to it.
The alternative — a bare machine and a root password — is how servers end up three years later with nobody sure what is running on them.
When we are not the right fit
- You want the cheapest instance you can find and will configure it yourself. That is a reasonable choice, and there is nothing we add to it.
- You need a large elastic fleet that scales with demand hour by hour. That is a cloud architecture question, and it should be designed as one.
- You want hardware shipped to your own premises. We specify and configure servers we can reach; on-site installation is a different arrangement.
What "delivered configured" actually contains
The phrase is used loosely, so here is the list. When a server is handed over, these are done and documented rather than left for you to discover.
- Operating system installed and updated to the current patch level on the day of handover, not the level of the install image.
- Firewall closed by default, with only the ports your workload actually needs opened, each one written down with a reason.
- SSH hardened: key-based login, password authentication disabled, root login restricted.
- Users and access created per person rather than one shared account, so an access review is possible later.
- Backups scheduled and a first restore tested before handover, not after.
- Monitoring connected for availability, disk, memory and certificate expiry.
- Timezone, locale and NTP set correctly, because log timestamps that drift make every future investigation harder.
- A written inventory of what is installed, what is listening, and where the backups go.
Storage, RAM and the two mistakes people make
Scoping goes wrong in the same two places almost every time.
Storage is sized for today's data rather than the growth curve. A database that grew 40% last year will not stop because a server was bought. The number that matters is not the current size but the size in eighteen months, plus the headroom the backup process needs while it runs.
RAM is sized for average load rather than peak. A server that is comfortable at the daily average and swapping at the weekly peak is a server that feels fine in testing and fails in production. We ask about peaks, not averages, and about what happens during the backup window — which is frequently the busiest the disk gets all day.
The output of scoping is a written specification: what is being bought, what it is sized for, and what it will not cover. If you take that document to another supplier and buy there instead, it still holds — that is what makes it worth writing.
Frequently asked questions
Do I need a dedicated server or is a VPS enough?
For most business websites, a VPS is enough and the difference is not visible to visitors. Dedicated hardware earns its cost when you have sustained disk I/O, software licensed per physical core, or a rule that the machine cannot be shared. We will tell you which of those applies to you before quoting.
What happens if we outgrow the server?
On a virtual server, growth is usually a resize with a short reboot. On dedicated hardware it means a migration to a bigger machine, which is why the scoping conversation asks about growth over the last year rather than only about today.
Is the operating system licence included?
Linux distributions we deploy are free to license. Where a workload needs a commercial operating system or a control panel, that licence is a separate line and it appears in the written specification before anything is ordered.
Can you move our existing setup onto the new server?
Yes. Migration follows the same parallel-run sequence as a hosting move: build and test the new machine while the old one keeps serving, lower the DNS time-to-live in advance, sync, then cut over in a low-traffic window.
Who has root access?
You do, if you want it. Some clients keep root and call us for specific work; others prefer we hold it and they hold the account. Either arrangement is fine as long as it is written down rather than assumed.
Do you manage the server after it is delivered?
Only if you ask. Delivery includes the initial configuration and documentation; ongoing monitoring, patching and backup verification are a separate service, and plenty of clients run the server themselves from day one.