Capacity planning
Choose an appropriately sized database compute for an application's workload.
Neon · Const
Compute units
Compute Units (CUs) measure the size and processing capacity of a Neon compute. Each CU provides approximately 4 GB of RAM plus associated CPU and local SSD resources.
How it works
Compute Units (CUs) express the size and processing capacity of a Neon compute. Each CU provides approximately 4 GB of RAM plus associated CPU and local SSD resources, so increasing the CU value scales these resources linearly.
Key points
- A compute's size is the number of CUs assigned to it.
- Higher CU values provide greater processing capacity.
- Every CU contributes approximately 4 GB of RAM plus CPU and local SSD capacity.
- Resource allocation grows linearly as the compute scales up.
- On the Free plan, a compute can reach 2 CU, corresponding to approximately 8 GB of RAM.
Common mistakes
- Choosing a larger compute increases compute-hour consumption for the same amount of active runtime.
Example: Estimate RAM from compute units
const computeUnits = 1;
const ramGb = computeUnits * 4;
console.log(`${computeUnits} CU provides approximately ${ramGb} GB RAM`);Calculates the approximate RAM associated with a one-CU compute.
When to use it
Vertical scaling
Increase compute capacity when a workload needs more processing power or memory.
Usage estimation
Estimate compute usage by combining a compute's CU size with its active runtime.
Interview questions
QWhat does one Compute Unit represent in Neon?
A Compute Unit is Neon's measure of compute size and processing capacity. One CU provides approximately 4 GB of RAM, together with associated CPU and local SSD resources.
Follow-up: How does increasing the CU value affect resources? It scales the allocated RAM, CPU, and local SSD resources linearly.
QWhat CU sizes can a Neon compute have?
A compute's size is expressed as a CU value, with current supported fixed sizes ranging from 0.25 to 56 CU. The available range depends on the plan.
Follow-up: What is the practical resource estimate for 8 CU? Approximately 32 GB of RAM, since each CU provides about 4 GB.
QHow are Compute Units related to billing usage?
CU-hours measure compute usage by combining compute size with active runtime. A compute running at 2 CU accumulates two CU-hours for each hour it remains active.
Follow-up: How can usage be reduced without changing workload duration? Use a smaller compute when its processing capacity and memory are sufficient.
Related
Compute sizeCompute Unit (CU)CU-hoursAutoscalingFixed size