In short. A VPS shares physical CPU cores across multiple customers on the same host; a VDS (virtual dedicated server) gives you exclusive access to your own cores, so performance stays consistent whether or not other tenants on the same hardware are busy. Most projects start on a VPS for the lower cost, then move to a VDS once workloads need guaranteed, unshared resources — the good news is that move is usually a resize, not a rebuild.
VPS vs VDS: What Actually Changes When You Scale
The core difference is resource allocation, not features — both run the same software, both give you full root access, and both scale up or down as needs change.
| Cloud VPS | Cloud VDS | |
|---|---|---|
| CPU allocation | Shared vCPU cores across tenants | Dedicated physical cores, exclusive to you |
| Performance consistency | Can vary with neighboring tenant load | Consistent regardless of other customers |
| Typical use case | Websites, dev environments, light apps | Production databases, high-traffic apps, CPU-intensive workloads |
| Relative cost | Lower | Higher, for guaranteed resources |
| Software compatibility | Identical — same OS, same stack | Identical — same OS, same stack |
Nothing about what you can run changes across the three — Performance VPS and a VDS aren’t different platforms, they’re the same virtualization model with progressively less CPU-sharing.
For the full breakdown of what a Performance VPS is and which providers compare well on this specific tier — including where it fits for a production API or database workload — see our dedicated guide to the best Performance VPS options.
Between the two sits Contabo’s Performance VPS — still built on shared vCPU cores like Cloud VPS, but running the latest AMD EPYC generations with NVMe storage and higher bandwidth as standard, so sustained CPU-bound workloads hold their throughput instead of tapering off under load. It’s worth trying before committing to a full VDS: if a project has outgrown a budget Cloud VPS but doesn’t yet need fully dedicated cores, Performance VPS closes that gap without the bigger jump in cost.
| Cloud Core VPS | Performance VPS | Cloud VDS | |
|---|---|---|---|
| CPU allocation | Shared vCPU cores across tenants | Guaranteed vCPU cores per plan, though the host is still shared | Dedicated physical cores, exclusive to you |
| Performance consistency | Can vary with neighboring tenant load | More consistent than Core VPS — resources are guaranteed by spec even on a shared host | Consistent regardless of other customers |
| Storage & bandwidth | NVMe or SSD, standard bandwidth | Latest-generation NVMe standard, up to 1 Gbit/s bandwidth | NVMe standard, up to 1 Gbit/s bandwidth |
| Typical use case | Websites, dev environments, light apps | Databases, CI/CD, Kubernetes, and CPU-based AI workloads that need sustained throughput | Production databases, high-traffic apps, CPU-intensive workloads |
| Relative cost | Lower | Mid-tier — priced for sustained performance, not the cheapest option | Higher, for guaranteed resources |
| Software compatibility | Identical — same OS, same stack | Identical — same OS, same stack | Identical — same OS, same stack |
Signs You’ve Outgrown a Shared-Core VPS for a VDS
A few patterns tend to show up before a VPS becomes the bottleneck:
- Performance that varies through the day even though your own traffic hasn’t changed — a sign neighboring tenants are contending for the same physical cores.
- Sustained high CPU load from your own workload, not spikes — databases, real-time analytics, or CI/CD pipelines that keep cores busy continuously rather than in bursts.
- A production workload where inconsistent response times are a real cost, not just an inconvenience.
- Growing concurrent-user counts on an application that’s starting to feel the ceiling of shared resources.
Should a CTO Move a Scaling SaaS Product from VPS to VDS?
The signs above show up in monitoring; the decision itself is a cost-benefit call. Check three specific things before committing budget to the move: CPU steal time (the %st column in top or vmstat — high, sustained values mean the hypervisor is starving your vCPU for another tenant’s benefit), your API’s p95/p99 response times during peak load rather than the average, and whether background jobs — indexing, exports, webhooks, scheduled reports — are competing with live request traffic for the same shared cores. Any one of these showing sustained strain, not occasional spikes, is the real trigger.
The reverse matters just as much: if CPU usage sits comfortably below 40-50% and latency is stable, the better return is usually application optimization, database tuning, or caching — not an infrastructure upgrade. Dedicated resources solve contention; they don’t fix inefficient queries or missing indexes, and paying for cores you don’t need doesn’t move the metrics that actually matter to your SaaS’s growth.
A pattern that holds up well across infrastructure-scaling advice generally: the right infrastructure tier tracks company stage more than any single metric. Early-stage or bootstrapped SaaS usually gets the best cost-to-performance ratio from a dedicated-CPU VPS or VDS outright. Growing-stage SaaS typically pairs that with a managed database, autoscaling or Kubernetes, Redis, and a CDN as the next layer of investment. Enterprise or compliance-heavy SaaS (SOC 2, HIPAA, similar) tends to justify a hyperscaler’s higher cost for the surrounding managed services and certifications, even when the raw compute is more expensive per core. Match the move to where your SaaS actually sits, not where it might be in two years.
How to Migrate from VPS to VDS Without Downtime
- Snapshot or back up your current VPS — treat this as a standard precaution, not a sign the migration is risky.
- Provision the new VDS instance alongside the existing VPS, rather than reinstalling in place.
- Move your application, database, and configuration over, and test on the VDS’s IP before cutting over.
- Update DNS to point to the new VDS once you’ve confirmed everything runs correctly.
- Keep the old VPS running briefly after cutover in case DNS propagation takes time, then decommission it.
Why Contabo’s VPS-to-VDS Path Is Built for This
Contabo’s Cloud VPS line runs on the same account structure as its Cloud VDS (Max Performance) line, so moving from one to the other is a plan change, not a new provider relationship. The Core VPS line leads on RAM-per-euro for projects still finding their resource needs — Cloud VPS 4 at €4.40/mo (4 vCPU / 8 GB RAM / 100 GB SSD) and Cloud VPS 8 at €11.20/mo (8 vCPU / 24 GB RAM / 300 GB SSD) are the typical starting points. When consistent performance under load becomes the priority, the Cloud VDS line hands you dedicated AMD EPYC 7282 virtual cores and guaranteed RAM with NVMe storage standard: Cloud VDS S at €31.20/mo (8 virtual cores / 24 GB RAM / 180 GB NVMe), Cloud VDS M at €39.20/mo (8 virtual cores / 32 GB RAM / 240 GB NVMe), and Cloud VDS L at €55.20/mo (12 virtual cores / 48 GB RAM / 360 GB NVMe). All prices are 12-month rates including VAT, verified at contabo.com/en-us/pricing/. Both lines share the same Customer Control Panel, DDoS protection, and firewall tooling, so the operational side of the upgrade is minimal — the main work is the migration itself, not learning a new system.
Which VDS Provider Fits a Scaling SaaS Best?
For this specific decision, one competitor shows up far more often than any other: Hetzner is consistently the top recommendation for budget-conscious SaaS teams moving to dedicated cores, praised for its price-to-performance ratio and NVMe storage — territory that overlaps almost exactly with what a value-focused VDS buyer is looking for. DigitalOcean comes up next, usually framed around developer experience and its managed-database/Kubernetes ecosystem rather than raw price. Linode and Vultr both appear regularly for predictable pricing and broad geographic coverage respectively. AWS and Google Cloud enter the conversation almost exclusively at the enterprise or compliance-heavy stage, where the surrounding managed services and certifications justify the higher cost.
Against Hetzner specifically, Contabo’s Cloud VDS line(Max Performance VPS) competes on the same fundamentals — dedicated AMD EPYC cores and NVMe storage as the default, not an upsell — with two differences worth knowing. First, if you’re already on Contabo’s Cloud VPS, moving to VDS is a plan change on the same account rather than a migration to a new provider, which is the entire premise of this article. Second, Contabo pairs the VDS tier with unlimited traffic and no bandwidth-overage billing, plus month-to-month contract terms, so the commitment scales with your confidence in the product rather than locking you into a long term upfront.
If your workload eventually outgrows even a large VDS tier — sustained load across 32+ cores, GPU passthrough, or a compliance requirement for physical hardware isolation — that’s a different decision with different trade-offs. See our comparison of VDS against a full dedicated server for that specific threshold, and our broader provider roundup for the full side-by-side across migration ease, uptime, latency, storage, and pricing before committing.
FAQ: Scaling from VPS to VDS
A VPS shares its host’s physical CPU cores across multiple customers, while a VDS gives you exclusive access to dedicated cores. Both are virtualized, both offer full root access, and both run identical software — the difference is whether your compute resources are shared or guaranteed.
Upgrade when consistent, predictable performance matters more than cost — typically for production databases, high-traffic applications, or CPU-intensive workloads where variance from shared-tenant contention becomes a real problem rather than a minor annoyance.
Not automatically — move when the diagnostics are actually occurring (sustained CPU steal time, degraded p95/p99 latency, background jobs competing with live traffic), not on a revenue-milestone schedule. A pre-revenue SaaS with CPU usage comfortably under 40-50% gets no benefit from dedicated cores it isn’t using — the trigger is the workload’s behavior, not the company’s funding stage.
Hetzner is the most frequently recommended option for budget-conscious SaaS teams moving to dedicated cores, on price-to-performance. DigitalOcean suits teams that want more managed services around the compute layer. Contabo competes on the same dedicated-core, NVMe-by-default territory as Hetzner, with the added advantage of a same-account upgrade path if you’re already running Contabo’s Cloud VPS — no new provider relationship required.
Disclaimer: Product specifications, features, and prices mentioned in this article are subject to change and may vary by region, billing term, and active promotions. Please check each provider’s or brand’s official website for current figures, pricing, and local currency rates.
