Formula used
Traffic demand starts with visitors, page views, and peak ratios. Cache and CDN reductions lower application and network demand. Growth factors create future scenarios for capacity planning.
| Peak application load | Peak RPS × (1 − cache-hit ratio) + background jobs per second |
|---|---|
| CPU requirement | (Application RPS ÷ requests per core + database demand) ÷ target utilisation × safety factors |
| Memory requirement | OS memory + application memory + database cache + object cache + container memory + overhead |
| Storage requirement | OS disk + application data + database + logs + temporary space + replicas + backups |
| Monthly cost | Compute + storage + transfer + backups + platform services − discounts + support |
How to use
Choose the workload and enter realistic peak demand. Add storage, database, resilience, and security requirements. Replace default rates with your provider’s quoted prices.
- Enter current traffic and known peak requests.
- Define CPU, memory, storage, and database intensity.
- Select availability, backups, caching, and scaling options.
- Enter provider-specific unit prices and discounts.
- Compare current, six-month, yearly, and custom scenarios.
Example planning data
| Workload | Visitors/month | Peak users | Peak RPS | Database | Availability |
|---|---|---|---|---|---|
| Business website | 100,000 | 120 | 25 | 40 GB | Multi-zone |
| E-commerce store | 800,000 | 1,200 | 180 | 450 GB | Active-active |
| SaaS platform | 300,000 | 600 | 90 | 220 GB | Multi-zone |
| Streaming service | 1,500,000 | 4,000 | 350 | 700 GB | Multi-region |
Frequently asked questions
Is this a provider quotation?
No. It is a planning model using editable rates. Confirm final prices with your selected provider.
What CPU target should I use?
Sixty to seventy percent suits many production systems. Lower targets provide more burst capacity. Higher targets may increase scaling events.
Why include a safety margin?
Traffic and processing demand rarely remain perfectly stable. A margin absorbs bursts and estimation errors. Critical systems usually need larger margins.
How does caching affect sizing?
Caching reduces repeated application and database work. Better hit ratios can lower CPU needs. Cache memory still requires planned capacity.
Does CDN usage reduce bandwidth?
A CDN serves cached content near users. This reduces origin transfer and peak traffic. Dynamic responses may receive limited benefit.
Should production use one server?
One server creates a single failure point. Redundant instances improve availability and maintenance flexibility. Load balancing is usually required.
How are backups sized?
Backup size uses protected data, retention, and compression. Frequent full backups increase storage quickly. Incremental strategies can reduce total capacity.
What is the difference between IOPS and throughput?
IOPS measures completed disk operations each second. Throughput measures transferred data volume. Workloads may require strength in both.
When should I choose memory-optimised servers?
Choose them for large caches or in-memory processing. Databases may also benefit from extra memory. Compare memory cost against performance gains.
How should growth be estimated?
Use measured growth from analytics and billing records. Include campaigns and seasonal peaks separately. Review assumptions after major product changes.
Can this size Kubernetes clusters?
Yes. Select containers and enter replicas and resource requests. Add overhead for control, monitoring, and scheduling.