01 · Cloud Computing
Server solution design and build
We design and build environments for specific applications: ERP and B2B systems, e-commerce platforms, databases and AI. Sizing backed by calculations, architecture documented end to end, performance and failover tested before acceptance — and maintenance afterwards if you want it.
Who it's for
Who uses this service
- Companies deploying a new system (ERP, B2B, e-commerce) that needs an environment built for it
- Software houses needing an infrastructure partner for their clients
- Organizations migrating applications off legacy servers
When it makes sense
Signs that the time is right
- The application vendor provided hardware requirements, but nobody knows how to meet them
- The application is slow, and nobody knows whether the infrastructure is to blame
- The planned system has availability requirements the current environment can't meet
What we build
Environments designed for one specific system
We start from the application and its requirements, not from a price list of servers. The result is an environment that is fast enough, available enough and no more expensive than it has to be.
Environments built to the software vendor's requirements, with the performance, availability and backup policy the business case demands.
Application, database and cache layers separated, sized for seasonal peaks and campaign traffic, with a CDN and WAF in front.
PostgreSQL, MySQL/MariaDB and MS SQL on NVMe with replication, tuned for transactional or analytical workloads.
Servers and clusters for model inference and training, with fast storage and network fabric — see GPU servers.
Proxmox or VMware clusters with shared storage, so many systems run on one well-designed platform.
Environments that survive the loss of a node, a rack or a location — see High Availability.
Architecture
Every layer designed, not assumed
A system is only as good as its weakest layer. We design all of them together and document why each decision was made.
Physical or virtual servers, CPU generation and core count, memory, GPU where the workload needs it — sized from real requirements.
NVMe, SSD or hybrid, ZFS, Ceph or an array, with the IOPS and capacity headroom the application will need in three years.
Segmentation, redundant links, load balancing and — for clusters — separate storage and management networks. See network engineering.
Operating systems, databases, application and web servers, containers, queues and caches, configured for performance instead of defaults.
Hardening, access control, firewall policy, backup and recovery plans built in from the first diagram, not added later.
Monitoring, alerting, log collection, update policy and runbooks — so the environment can be run by people who did not build it.
Sizing
How we size an environment
Sizing is arithmetic, not intuition. We collect the inputs, calculate the requirements and then verify them under load before you sign off.
| What we ask about | What it decides | How we verify it |
|---|---|---|
| Users and concurrency | CPU cores, memory, application server count | Load tests with realistic scenarios |
| Transactions and peaks | Database resources, cache layer, queueing | Peak-traffic simulation |
| Data volume and growth | Storage capacity, tiering, retention | Growth projection for 3–5 years |
| IOPS and latency | Drive type, RAID or erasure coding, controllers | Storage benchmarks on the target hardware |
| Availability (RTO / RPO) | Redundancy level, clustering, replication, backup | Failover tests and restore tests |
| Integrations and traffic | Network bandwidth, links, firewall throughput | Throughput tests between components |
| Regulations and audits | Data location, logging, segmentation, documentation | Compliance checklist (GDPR, NIS2, DORA) |
| Vendor requirements | Supported versions and configurations | Written confirmation of compliance |
Every project ends with a sizing document you can show to the application vendor, your board or an auditor.
Where it runs
Our data center, your server room, cloud or hybrid
We compare the options honestly, with a 3–5 year total cost of ownership — including the costs most comparisons forget, such as licences, power, replacement hardware and staff time.
| Option | Strengths | Watch out for | Typically for |
|---|---|---|---|
| Our data center | Tier III, redundant power and links, hardware and operations handled by us | Requires good connectivity to your sites | Most business systems and e-commerce |
| Your server room | Full physical control, data on site | Power, cooling, physical security and spares are on you | Production plants, OT, strict internal policies |
| Public cloud | Elastic capacity, managed services, fast start | Costs grow with traffic and data transfer | Variable workloads, projects with unknown scale |
| Hybrid | Steady load on fixed cost, peaks in the cloud | Two environments to integrate and monitor | Systems with seasonal peaks or DR in a second location |
Hardware procurement, licences and vendor support contracts can be handled by us as part of the project — colocation and dedicated servers included.
Deliverables
What lands on your desk
Design work is only finished when somebody else can build, run and audit the environment from the documents.
Target architecture with diagrams and the reasoning behind every significant decision.
Configuration-level detail: addressing, VLANs, disks, volumes, users, parameters.
Calculations behind the resources and a bill of materials with hardware, licences and support contracts.
3–5 year cost comparison of the considered options, including licences, power and maintenance.
Performance and failover scenarios with results — the basis for formal acceptance.
Day-to-day procedures, recovery steps and contacts, ready for the team that will operate the system.
Technologies we build with
Scope
What the service includes
Discussion with the application vendor, load, availability, security, regulations.
Servers, storage, network, virtualization, backup — sized to the application, not oversized.
Our DC, your server room, public cloud or hybrid — with a 3–5 year TCO.
Installation of operating systems, databases, application servers, integrations.
Performance and failover testing before acceptance — with a formal test report.
Documentation, runbook, training or SLA-backed maintenance.
Process
How we work — four steps
Analysis (1–2 weeks)
Application and business requirements; constraints; budget.
Design
Architecture, cost estimate, schedule; alignment with the application vendor.
Build and migration
Environment, data migration, parallel testing.
Acceptance
Testing, documentation, handover or maintenance.
Outcome
What you get
- Documented architecture (HLD and LLD) with decision rationale
- A built and tested environment
- Performance and failover test reports
- Runbook, DR plan and a bill of materials
- Optional SLA-backed maintenance
Questions
Frequently asked questions
Do you also build the application itself?
We are not a software house — we build the environment, integrations and automations (for example in n8n). We work with your application vendor as the infrastructure partner, and we are used to being the ones who ask them the awkward questions about requirements.
Our data center or the public cloud?
It depends on the application, the regulations and the cost profile. We prepare a 3–5 year TCO for the realistic options — our data center, your server room, public cloud or hybrid — and recommend one, with the reasoning written down.
What if the application vendor requires a specific configuration?
We meet the vendor's requirements and document the compliance, which protects your support and warranty. Where a requirement is unrealistic or outdated, we say so and agree an alternative with the vendor in writing.
How long does the design phase take?
Analysis usually takes one to two weeks, after which you receive the architecture, cost estimate and schedule. The build depends on hardware lead times and the complexity of the migration — both are in the proposal.
Do you supply the hardware and licences?
Yes. We prepare the bill of materials and can handle procurement, licences and vendor support contracts, or you can buy everything yourself using our specification — whichever is better for your budget process.
Can you take over and fix an existing environment?
Yes. We start with an audit: what runs where, how it performs and where the risks are. Often a well-targeted redesign of storage, database configuration or network is cheaper than replacing everything.
How do you migrate without a long downtime?
We build the new environment in parallel, synchronize data, test it under load and switch over during an agreed window with a rollback plan ready. For most systems the visible interruption is measured in minutes.
How do you prove the environment really performs?
With tests: load and performance scenarios for the application, failover tests for redundancy and restore tests for backup. The results form a test report, which is the basis for acceptance.
Who runs the environment afterwards?
You can take it over with the documentation and runbook, or leave it to us under an SLA — see server administration. Many clients split it: their team handles the application, ours the infrastructure.
How is the project priced?
The design phase is a fixed price agreed after the first call. The build is priced per project, based on the approved architecture and bill of materials. Maintenance afterwards is a monthly fee for the agreed scope.
Related services
Often combined with this service
First step
Let's talk about server solution design and build.
30 minutes, no slide deck. We'll tell you straight whether this service solves your problem, what scope makes sense and how much it costs.
A proposal with scope and pricing within 48 hours of the call.
Go to contact