OVH: What It Actually Is and How to Navigate It
OVH, now rebranded as OVHcloud, started as a small French hosting company and grew into one of the largest publicly traded infrastructure providers in Europe. They make money by selling bare metal servers, virtual private servers, cloud instances, and colocation space. Their scale is real. They operate massive datacenters in Europe, North America, and Asia. That matters because it affects your latency and your compliance options if you're dealing with EU data residency requirements. Their cloud platform is built on a KVM hypervisor layer. Unlike some providers who wrap everything in proprietary abstractions, OVH's cloud instances are relatively straightforward virtual machines. You pick a region, choose a compute tier, attach storage, and boot. The interface is functional but not polished. I've been managing OVH infrastructure since before the rebrand happened, and the core concepts haven't changed much. Here's something most beginners miss about their cloud offering. OVH prices their instances in hourly bursts with a 100-hour minimum charge per month. That means if you spin up an instance for just a few hours, you still get billed for 100 hours. I learned this the hard way when I was testing a short-lived migration script. Spun up a few high-memory instances, ran the job, terminated them. The invoice came in at nearly the same cost as leaving them running all month. The workaround was to use their preemption (spot) instances instead. Those are significantly cheaper, though they can be terminated with only a two-minute warning if OVH needs that capacity back. For batch processing and non-critical workloads, preemption instances saved me roughly 60 to 70 percent compared to standard pricing.
Another counter-intuitive thing about OVH's infrastructure is their networking model. They provide DDoS protection by default on all their IP addresses, including cloud instances. This is unusual. Most providers charge extra for network-level DDoS mitigation. OVH includes it because they route traffic through massive scrubbing centers. The tradeoff is that their default firewall rules are somewhat aggressive. If you run applications that depend on certain ICMP types or specific port ranges for health checks, the default DDoS filter might drop your traffic. I encountered this when setting up a monitoring solution that relied on ICMP echo requests across multiple OVH cloud nodes. The monitoring server couldn't reach the instances. The fix was logging into the OVH control panel, navigating to the DDoS Protection section for each IP, and adding specific allow rules for the ICMP types my monitoring tool required. It took about ten minutes once I figured out where the setting lived, but finding it required digging through their documentation. Let me talk about their storage because this is where people get burned. OVH offers two main storage options for cloud instances: ephemeral storage and block storage. Ephemeral storage is tied to the hypervisor node your instance runs on. If the instance migrates for any reason, you lose everything on that disk. Block storage is persistent and survives migrations, but it costs extra. I've seen people deploy production databases on ephemeral disks expecting data to survive reboots. It doesn't always survive migrations. The rule of thumb is simple. Never store anything persistent on ephemeral storage. Attach a separate block storage volume for databases, logs, and application state. Their bare metal line is probably what they're best known for. If you need raw performance without the overhead of virtualization, their dedicated servers are competitive on price. The catch is that provisioning can take anywhere from an hour to several days depending on availability in your chosen datacenter. I once needed a server urgently and waited three days for the rack space and hardware to become available in Gravelines. If speed matters, check their stock status page before ordering, though even that isn't always perfectly accurate.
One more practical detail about their virtualization stack. OVH uses their own proprietary hypervisor called Nova, not standard OpenStack. This means some of the commands and tools you'd use with a standard OpenStack deployment won't work the same way. The API is similar but not identical. If you're automating infrastructure, write your scripts against the OVH API rather than assuming standard OpenStack compatibility. I spent a half day debugging Terraform scripts that worked perfectly against a standard OpenStack environment but failed against OVH because of differences in how they handle network creation and floating IP association. On the billing side, OVH is prepaid. You fund your account and services draw from your balance. There's no post-billing cycle. This is intentional. They want you to manage your own budget and stop services before the balance runs dry. It sounds simple but it's easy to overlook when you're juggling multiple projects. I had a project that accumulated enough unused instances to drain the account balance on a Friday evening. By Monday morning, everything was offline. The recovery wasn't terrible, but it took four hours of restoration work and lost access time for the team. Support quality varies depending on which plan you're on. Free support through their ticket system is slow. I typically get responses within 24 to 48 hours for non-critical issues. Paid support tiers exist and are meaningfully faster, but even then, their support staff tends to focus on infrastructure-level problems rather than application-level troubleshooting. They'll tell you if your network is down or your disk is full. They won't help you debug why your application is crashing. That's standard for this class of provider but worth stating explicitly.
Get the Full Details
If you need something more managed or if your workload requires tight integration with major cloud ecosystems, OVH isn't the best fit. They don't offer the same breadth of managed services as AWS or Azure. No managed Kubernetes that competes with EKS or AKS. No mature serverless offering. They have Kubernetes support through their managed KOA product but it's newer and less feature-rich than the major competitors. For pure compute, storage, and networking at competitive prices, OVH is solid. For anything requiring a wide ecosystem of managed services, you're better off elsewhere or using them as a secondary infrastructure layer rather than your primary one.
