Shared CPU and dedicated CPU plans can have similar-looking vCPU counts while delivering different performance characteristics. Shared plans are designed for workloads that burst and idle. Dedicated plans cost more to provide more predictable compute under sustained load.
The correct choice depends on CPU behavior, not whether the server is labelled production.
Key takeaways
- Shared CPU is best value for bursty and low-to-medium CPU workloads.
- Dedicated CPU is useful for sustained compute and predictable latency.
- Websites, development, VPNs, and small APIs usually fit shared CPU.
- Game servers, encoding, CI, analytics, and busy databases may justify dedicated CPU.
- Measure before upgrading.
How shared CPU works in practice
Shared CPU virtual machines run on host processors used by multiple tenants. Providers schedule workloads so each VM receives CPU time, but sustained performance can vary depending on the platform and neighbors.
This model is efficient and inexpensive for services that spend significant time waiting on users, networks, or databases.
What dedicated CPU buys
Dedicated CPU plans reserve processor resources more predictably for the instance, reducing contention. That makes performance easier to reason about for continuous compute workloads.
The premium is wasted if the application is idle most of the day.
Workload examples
Shared CPU is a sensible default for WordPress, small Docker stacks, VPNs, APIs, staging, monitoring, and personal services. Dedicated CPU is more attractive for build servers, video processing, game servers, busy databases, analytics, and sustained application traffic.
Some workloads need faster single-core performance rather than more cores, so inspect the CPU family as well as the allocation model.
A simple upgrade rule
Start with shared CPU unless the workload is known to be compute-intensive. Monitor utilization and response latency under realistic load.
Move to dedicated CPU when sustained compute saturation or variability is the measured problem. Do not use dedicated CPU as a substitute for fixing slow queries or inadequate memory.
Frequently asked questions
Is shared CPU bad for production?
No. Many production websites and applications are bursty and run very well on shared CPU.
When should I use dedicated CPU?
When sustained compute, predictable performance, or latency consistency materially affects the workload.
Does dedicated CPU mean a physical server?
Not necessarily. It usually means dedicated processor resources within a virtualized platform, not an entire physical machine.
Is dedicated CPU worth it for Docker?
Only when the containers perform enough sustained compute to justify the cost.