Redundant at every level, in two cities.

CACloud runs on redundant, enterprise-grade server hardware with replicated storage: every block of data is written to several disks on several servers, so no single disk or server holds the only copy. A failed drive is rebuilt from the other copies without a ticket.

The whole platform runs on one KVM-based virtualization stack, in both cities. A Private Cloud gives you your own nodes on that same stack, so a workload that must stay on isolated hardware can. The technology page lists what each product line uses.

Customers get the benefits of a hosted cloud along with Canada's privacy and governance rules, and with the same experience in both cities: the platform is standardised, so a server in Toronto and a server in Vancouver behave the same way.

The virtualization layer supports four ways to deploy, on the same platform.

Public cloud is sold as SSD Cloud Servers and Enterprise VM. Private cloud is the Private Cloud line. Ask us about hybrid and VPS layouts.

Enterprise VM and production Private Cloud clusters are highly available. If a host stops responding because its hardware failed, the platform restarts the affected machines on another host automatically, usually within minutes. You do not open a ticket first. SSD Cloud Servers are single-host machines: they get the same replicated storage, not automatic host failover.

That design backs a written commitment: 99.9% monthly availability for SSD Cloud Servers and 99.95% for production private clouds of three nodes or more, measured per region, with credits of up to 30% when we miss. The full terms are in the service level agreement.

Capacity is added continually, and load balancing spreads work across it, so there is always another host ready to pick up the slack when one is busy or down.

Resizing a server is a portal action. Sign in, change the CPU, memory or disk allocation, and reboot; the new resources are live in under a minute. There is no migration and no ticket. Load-balancing and smart resource scheduling decide where the resized machine runs.

Servers can be upgraded or downgraded in real time, so you can start small, test, and grow the same machine into production.

Each stack is redundant at every level, and its storage is replicated: every write lands on more than one disk and more than one server, so losing a drive, or a whole storage server, loses no data. Nothing your server needs sits on a single disk, or a single machine. The storage is network-attached and built on SSDs, so provisioning and upgrades are fast.

Backups, where enabled, are stored separately from the server they protect. Automatic backups run on a schedule; on-demand snapshots let you capture a machine before a risky change.

Each site reaches the internet over multiple redundant links to Tier 1 upstream providers, on a Canadian network backbone. Every server has a direct connection to that backbone, and traffic is not throttled.

Security is built into the same layer: intrusion prevention and firewall management sit in front of every host, and each customer gets a private VLAN, described on the security architecture page.

DDoS protection is Cloudflare Magic Transit, and it is always on at the network edge. We put it in after a run of ransom DDoS attacks, and it covers every server on the network — there is nothing to turn on and nothing to order.

Canadian Web Hosting, the company that operates CACloud, is Canadian-owned and has run hosting since 1998, with data centres in Toronto and Vancouver. Every service and location sits inside the group's SOC 2 Type II report; the compliance page explains what that covers and what Canadian residency means in practice.