Scale 4. More headroom to work with.

More allocation for measured overlap

Compare Scale 4 when measurements show CPU contention or memory pressure across multiple active services. A larger starting budget can accommodate different concurrency choices, but it does not repair an unbounded export, a slow external dependency or an incompatible migration.

When to choose a different shape

One larger host still has one host-level failure boundary. Consider separate components when maintenance, access control or recovery needs diverge, even if the combined resource total could fit.

Understand the option you are changing

CPU, memory, working storage, monthly transfer and IP choices solve different constraints. Backups are a separately priced option; extra disk alone is not independent recovery. Compare all recurring options before increasing the budget.

Check the next release as well as today’s traffic

Record artifact, schema changes, worker behavior and temporary space. Use the relevant operating guide to choose a measurable next step.

When this profile makes sense

Consider this when measurements point to CPU contention or memory pressure. Extra resources do not replace profiling.

Use a representative workload to validate memory peaks, disk needs, and CPU demand. No benchmark, traffic capacity, or guaranteed application performance is claimed.

What you manage

Choose the Linux image, install and update your application, and keep your own access, monitoring and recovery process. Actual supplier images and availability still need confirmation.