Cloud server pricing is the recurring cost of compute plus the services required to run, protect, and operate a workload. On Raff, General Purpose Linux VMs currently start at $4.99/month for 1 vCPU, 1 GB RAM, and 25 GB NVMe SSD. A more typical shared-vCPU 2 vCPU / 4 GB / 80 GB plan is $17.99/month.
The VM plan is only the first line of a production budget. Storage growth, backups, Windows licensing, additional IPs, databases, non-production environments, and other services can change the real monthly cost. Raff General Purpose, CPU-Optimized and Windows VMs have a 3 Gbps public port; Cost-Optimized Linux VMs have a 1 Gbps public port. VM bandwidth is unmetered without a VM egress fee on these families, but other Raff products may have different traffic policies.
If you are building ongoing controls after choosing an initial architecture, use Cloud Budget Guardrails for Startups alongside this pricing guide.
Cloud server pricing has several cost layers
| Cost layer | What you are paying for | When it matters |
|---|---|---|
| Compute | vCPU and RAM assigned to the VM | Every workload |
| Base storage | Operating system, application, database, and local files | Every workload |
| Additional storage | Volumes or object storage beyond the VM disk | Growing data sets |
| Network | Egress policy, public IPs, and private networking | Public or multi-server workloads |
| Data protection | Snapshots, automated backups, retention, and recovery storage | Production data |
| Operating system | Windows Server or other licensed software | Licensed workloads |
| Supporting services | Managed databases, object storage, additional VMs, and orchestration | Separated architectures |
| Operations | Patching, monitoring, incident response, migration, and support | Production environments |
| Idle capacity | Unused development, staging, preview, and oversized resources | Teams with multiple environments |
A provider's plan table usually shows the compute line first. A useful budget covers the complete stack.
Current Raff VM pricing starts at $4.99 per month
Raff currently offers three Linux VM families, accessible through the Raff Linux VM product page, plus a separate Windows VM family. These are not interchangeable solely because the vCPU and RAM counts match: CPU allocation, hardware generation, disk layout, public-port speed and uptime terms also affect the fit.
| Raff VM family | Example shape | Standard monthly price | What the price represents |
|---|---|---|---|
| Cost-Optimized (Linux only) | 2 vCPU / 4 GB / 40 GB NVMe | $9.99 | Budget-oriented, previous-generation hardware |
| Cost-Optimized (Linux only) | 4 vCPU / 8 GB / 80 GB NVMe | $19.99 | More RAM and vCPU at a lower cost per shape, but a different service class |
| General Purpose (Linux) | 1 vCPU / 1 GB / 25 GB NVMe | $4.99 | Lowest standard monthly Linux VM entry price |
| General Purpose (Linux) | 2 vCPU / 4 GB / 80 GB NVMe | $17.99 | Shared-vCPU reference plan |
| CPU-Optimized (Linux) | 2 vCPU / 4 GB / 80 GB NVMe | $36.99 | Dedicated-vCPU reference plan |
| Windows | 1 vCPU / 2 GB / 50 GB NVMe | $28.99 total | $13.99 compute + $15.00 Windows Server Standard licence |
Cost-Optimized runs on previous-generation hardware with DDR4 memory, a 1 Gbps public port and local NVMe on the server's own node. It is Linux only and has no uptime SLA or SLA service credits. It is best considered for dev/test, side projects, recoverable batch workloads and other non-critical uses. Resize is offered between Cost-Optimized sizes; do not assume an in-place migration to the other VM families.
General Purpose uses shared vCPU and is suited to balanced or variable workloads. CPU-Optimized uses dedicated-vCPU allocation for sustained CPU demand. These families use DDR5 hardware and a 3 Gbps public port. They, along with Windows VMs, are covered by Raff's 99.90% monthly uptime SLA, subject to the published terms.
All three Linux VM families advertise unmetered VM bandwidth with no VM egress charge, but unmetered does not mean identical port speeds, performance or availability guarantees. Included VM storage, backup/snapshot free-pool rules, VPC, DDoS protection and optional add-ons should be checked for the selected plan on the live pricing page.
Comparing two providers by vCPU and RAM alone can therefore be misleading. Compare CPU allocation, hardware generation, local versus separate storage, disk size, support, transfer policy, location and SLA alongside the price.
CPU and RAM usually decide the VM tier first
CPU and memory normally determine the VM size before additional storage does.
More vCPUs provide capacity for concurrent requests, builds, background workers, report generation, and database queries. RAM supports application processes, database working sets, operating-system cache, containers, queues, caches, and concurrent Windows sessions.
Memory pressure can be more disruptive than moderate CPU pressure. Once a VM begins swapping heavily or terminating processes, a cheaper plan can create more engineering work than the price difference saves.
Use Choosing the Right VM Size before comparing monthly totals.
Storage pricing depends on where the data belongs
The disk included with a VM is only one storage layer. Production data often grows through databases, uploaded files, logs, container images, build artifacts, package caches, and backup staging.
Raff currently prices additional storage differently by purpose:
| Storage type | Current Raff price | Best fit |
|---|---|---|
| Block storage volume | $0.08/GB/month | Expandable disk-like storage for VMs and persistent workloads |
| Snapshot storage | $0.06/GB/month above the free pool | Point-in-time rollback |
| Automated backup storage | $0.06/GB/month above the free pool | Scheduled retained recovery |
| Object Storage | $7/month for first 100 GB | Backups, media, archives, and application objects |
Object Storage includes 1 TB of egress in the first 100 GB package; additional object-storage egress is billed at $0.01/GB. This is why a generic statement such as "Raff has no egress fees" is too broad: VM and Kubernetes node traffic is unmetered with no egress fee, while Object Storage, Apps, and Functions have their own egress rules.
Do not choose storage only by price per gigabyte. Access pattern, recovery time, application compatibility, and whether data needs disk semantics matter.
Bandwidth policy can change the real monthly bill
Providers commonly use one of three network pricing models:
- unmetered traffic;
- a monthly transfer allowance with overage;
- per-gigabyte egress billing.
This difference matters for public APIs, file downloads, media delivery, backup transfer, and high-traffic applications.
For Raff VMs, Cost-Optimized has a 1 Gbps public port, while General Purpose, CPU-Optimized and Windows have a 3 Gbps public port. VM traffic is unmetered with no VM egress fee in each of these families; port capacity and network performance still differ. VPC/private traffic is also free and unmetered. Reserved IP addresses are separate resources: reserved IPv4 is $3/month and reserved IPv6 is $1/month.
When comparing two VM providers, normalize both the compute shape and the network policy. A lower base VM price can lose its advantage when outbound traffic is charged separately.
Backups and snapshots are part of production cost
A production VM without a recovery plan is cheaper only until recovery is needed.
Raff uses a shared free pool for backup and snapshot storage. Additional snapshot and automated-backup storage is $0.06/GB/month. There is no fixed three-slot backup limit: automated backups can use Daily and/or Weekly schedules with configurable retention, and snapshots are not limited by a slot count.
Budget protection around the workload's Recovery Point Objective (RPO) and Recovery Time Objective (RTO), not around the lowest possible storage bill. Important databases may also need database-aware backups or point-in-time recovery in addition to VM-level protection.
Use RPO vs RTO for Cloud Backups before selecting retention.
Windows cloud server pricing has two separate line items
A production Windows Server workload can have a compute charge and a Windows licence charge.
Raff's current Windows VM family starts at $28.99/month including the Windows Server Standard licence for 1 vCPU, 2 GB RAM, and 50 GB NVMe SSD. The invoice separates this into $13.99/month compute + $15.00/month Windows Server Standard licence. Eligible BYOL is also supported under Microsoft's applicable outsourcing rights.
One pricing decision we made at Raff is to keep Windows compute and the Windows licence separate. That prevents an eligible BYOL customer from being charged twice for a licence they already have the right to use.
Remote Desktop Services can add another licensing layer. Real multi-user RDP beyond the two administrative sessions requires the RDS role and a User SAL, currently $8.00/month per user through Raff.
For Windows cost comparisons, lead with the licence-inclusive total, then check the compute/licence split, RDS requirements, SQL Server or other Microsoft software, storage, and backup requirements separately. Use the Windows VM product page and the current licensing guidance rather than treating the compute-only figure as the Windows entry price.
Supporting services change architecture cost
A single VM may be enough at launch. The monthly bill changes when reliability or separation requirements justify additional components.
| Requirement | Possible additional service |
|---|---|
| Separate application and database | Second VM or managed database |
| Higher application availability | Additional application capacity and traffic distribution |
| Private service communication | VPC/private networking and firewall rules |
| Expandable persistent disk | Block storage volume |
| User uploads or archive data | Object storage |
| Deployment isolation | Staging or preview environment |
| Managed database operations | Managed database service |
| Container orchestration | Kubernetes nodes and control-plane option |
The cost is justified when the service reduces a defined reliability, security, performance, or operating risk. Adding components without defining the problem they solve creates complexity and spend at the same time.
Managed services trade invoice cost for operational ownership
A self-managed database on a VM can have a lower visible infrastructure price. A managed database may shift backups, updates, monitoring, failover, or other operational responsibilities to the service layer.
| Area | Self-managed | Managed service |
|---|---|---|
| OS patching | Your team | Provider-managed service layer |
| Database updates | Your team | Depends on service policy |
| Backup design | Your team | Often integrated |
| Monitoring | Your team | Often integrated |
| Failover | Must be designed | May be available by tier |
| Direct monthly price | Often lower | Often higher |
| Engineering effort | Higher | Lower for covered operations |
The cheaper invoice is not always the cheaper operating model. Include engineering time and incident ownership in the decision.
Idle infrastructure creates silent monthly cost
Cloud waste often comes from resources that were useful when created but no longer have an owner.
Common examples include abandoned test VMs, oversized staging servers, preview environments left online, unused volumes, old snapshots, duplicate databases, and temporary migration servers.
Every non-production resource should have an owner, a purpose, and a review or expiry date. Idle Infrastructure Cost covers the keep, resize, schedule, archive, or delete decision in more detail.
Monthly, yearly, 24-month, and pay-as-you-go billing serve different needs
Raff supports subscriptions and pay-as-you-go usage.
- Monthly: the standard monthly plan price.
- Yearly: priced at 10 times the monthly rate, equivalent to two months free.
- 24-month: priced at 18 times the monthly rate.
- Pay as you go: hourly PAYG exists on VM rows and draws from prepaid account balance as resources run.
Longer commitments make sense when workload size is stable and the discount is worth more than the lost flexibility. Monthly or PAYG is safer when you are validating a workload, migrating, or expect resource needs to change.
This corrects an older version of this guide that described Raff's prepaid usage as something that should not be called hourly billing. The current product supports hourly PAYG, so provider comparisons should normalize the actual billing model used.
The real monthly cloud server cost is a simple sum
Use this model before comparing providers:
Monthly infrastructure cost = base compute + additional storage + backups and snapshots + operating-system and software licensing + supporting services + network charges that apply to the product + additional IP resources + idle resources
Then include the operating cost:
Total operating cost = monthly infrastructure cost + engineering time + maintenance + incident response + migration and recovery work
Not every team needs to convert engineering time into an exact dollar amount. It should still be part of the decision because a cheaper service can require substantially more operational ownership.
A cloud server pricing decision should follow the workload
Use this order when evaluating cloud server pricing:
- Define the workload. Record the application, users, data, traffic, and performance expectations.
- Choose the appropriate VM class and availability level. Cost-Optimized trades newer hardware, port speed and a contractual uptime commitment for a lower-cost resource option. Compare shared with shared and dedicated allocation with dedicated allocation; verify the selected class rather than assuming equal performance from equal vCPU counts.
- Choose the smallest safe VM. Include normal headroom, not speculative multi-year growth.
- Add persistent data costs. Include disk growth, volumes, objects, logs, and uploads.
- Add recovery. Define RPO, RTO, backup frequency, snapshot use, and retention.
- Normalize network pricing. Separate VM traffic rules from egress-charged products and competitor policies.
- Confirm licensing. Keep Windows Server, RDS, SQL Server, and application licensing separate from compute.
- Add only required services. Databases, additional environments, or orchestration should solve a known problem.
- Choose the billing term. Balance commitment savings against flexibility.
- Verify the live price. Recheck the final plan and add-ons before purchase.
