Infrastructure Planning: The Decision That Shapes Everything
Building web hosting infrastructure from scratch in 2026 requires making a series of interconnected decisions — compute platform, operating system, control panel, billing system, support tooling, and monitoring stack — where early choices constrain later options. This guide walks through the complete architecture of a production web hosting business, from server selection to customer-facing billing automation.
Bare Metal vs VPS vs Cloud
The compute platform decision is the most consequential. Each option suits different business stages and customer profiles:
Bare Metal dedicated servers (Hetzner, OVHcloud, PhoenixNAP) deliver the best price-to-performance ratio at scale. A Hetzner AX41 dedicated server — AMD Ryzen 5 3600, 64GB RAM, 2x512GB NVMe — costs approximately €40/month and can host 200–400 shared hosting accounts profitably. The trade-off is hardware failure risk, longer provisioning time (hours to days vs seconds), and requiring more operational maturity.
VPS hosting (DigitalOcean, Vultr, Linode, Hetzner Cloud) is ideal for starting out and scaling incrementally. Provision new servers in 60 seconds, scale vertically with a button click, and pay only for what you use. For a new hosting business targeting under 100 accounts, a $40/month 8GB RAM VPS is sufficient. The per-unit economics are worse than bare metal but operational simplicity and flexibility compensate during early growth.
Cloud platforms (AWS, GCP, Azure) add managed services — RDS for databases, S3 for backups, Route 53 for DNS, CloudFront for CDN — that reduce operational complexity at the cost of vendor lock-in and less predictable pricing. Cloud makes sense for hosting infrastructure with variable traffic patterns or where managed database and CDN services are core to the product differentiation.
WHM/cPanel Setup
Install cPanel/WHM on a fresh CentOS 7/AlmaLinux 8/CloudLinux 8 server:
curl -o latest -L https://securedownloads.cpanel.net/latest
sh latestThe installation takes 30–60 minutes. Post-installation, complete the WHM Setup Wizard: set the primary nameservers (ns1.yourdomain.com, ns2.yourdomain.com), configure the contact email, enable AutoSSL via Let's Encrypt, and set your default hosting packages. Install CloudLinux immediately after cPanel for resource isolation — the conversion script handles the OS layer migration without a reinstall.
Network Configuration
Configure reverse DNS (PTR records) for your server IP — critical for email deliverability. Contact your server provider to set PTR records to match your server's hostname. Configure firewall rules with CSF (ConfigServer Security & Firewall) — allow ports 80, 443, 25, 587, 465, 110, 995, 143, 993, 21, 22, 2082, 2083, 2086, 2087 and block everything else. Set up IP address blocks: a /29 (8 usable IPs) gives you flexibility to assign dedicated IPs to premium customers.
Storage and RAID
For bare metal servers, configure RAID-1 (mirroring) for the OS drive and RAID-10 (striped mirrors) for customer data if four drives are available. RAID-10 provides both redundancy and read performance. On VPS and cloud platforms, rely on provider-level redundancy (typically RAID-10 or erasure coding at the storage layer) and supplement with off-server backups rather than managing RAID manually. For NVMe storage, ZFS provides software-level RAID with checksumming that catches silent data corruption that hardware RAID misses — worth the additional configuration effort on storage-critical servers.
CDN Integration with Cloudflare
Integrate Cloudflare as the CDN and DNS provider for all customer domains. Add your server IP as an A record origin in Cloudflare and enable proxying. Cloudflare's free tier provides DDoS protection, global CDN caching, SSL termination, and bot management — services that would cost hundreds of dollars monthly from other providers. For hosting resellers, Cloudflare for SaaS allows you to provision Cloudflare's CDN for custom customer domains programmatically. Configure page rules or Cloudflare's Cache Rules to cache static assets for 30 days and bypass cache for dynamic content (WordPress admin, WooCommerce checkout).
WHMCS Billing System
Install WHMCS on a separate server or subdomain (billing.yourdomain.com) from your hosting infrastructure — keeping billing separate from customer hosting ensures that a compromised customer account cannot affect billing data. Configure WHMCS with your cPanel/WHM server using the cPanel Extended provisioning module. Set up:
- Product packages matching your WHM hosting plans
- Automated welcome emails and provisioning on payment
- Automated suspension on invoice overdue (after 3 days) and termination (after 30 days)
- PayPal and Stripe payment gateway modules
- Domain registrar integration (Enom, Namecheap, ResellerClub) for domain upsells
Support System Integration
Freshdesk's Growth plan at $15/agent/month provides ticketing, knowledge base, and email integration sufficient for hosting support. Integrate Freshdesk with WHMCS via the WHMCS-Freshdesk integration module so support agents can see customer account details, service status, and invoice history without leaving the Freshdesk interface. Configure SLA policies: 4-hour first response for technical issues, 30-minute response for billing. Set up automated ticket creation from cPanel's Backup Notification emails so disk-full and backup-failure alerts create support tickets automatically.
Monitoring Stack
Deploy a monitoring stack separate from your hosting servers:
- Uptime Kuma (self-hosted on a $5 VPS) for external HTTP monitoring of all customer domains and server management interfaces — alerting via Telegram or email when sites go down
- Netdata on each hosting server for real-time resource monitoring accessible via Netdata Cloud
- Prometheus + Grafana for long-term metrics aggregation and alerting rules (disk above 80%, memory above 85%, load average above CPU count)
The complete infrastructure stack described here — bare metal or VPS compute, CloudLinux + cPanel, Cloudflare CDN, WHMCS billing, Freshdesk support, and a layered monitoring setup — can be assembled and operational in a week. The investment in proper architecture at launch prevents the costly migrations and customer-impacting outages that come from outgrowing a hastily assembled infrastructure.