Aurora → Neon Scale Plan Pricing Question for Production Setup
Hi everyone!
I’m currently running a production-grade app on AWS Aurora, and I’m considering switching to Neon Scale plan. I’d like to understand the pricing implications based on my real-world usage. Here are my current specs on Aurora:
* Instance:
* CPU utilization: ~7% on average
* Disk usage: 1 GB (assume total size)
I want to migrate my main production database to Neon. Based on Neon Scale pricing:
* Compute is billed in CU-hours, where 1 CU = 1 vCPU + 4 GB RAM.
* Neon can autoscale down to 0.25 CU (0.25 vCPU + 1 GB RAM).
* My DB is mostly idle, so it would run near this minimum capacity most of the time.
* Neon Scale plan rate: $0.26 per CU-hour
* Storage: $0.35 per GB-month
* Egress / PITR and other fees are ignored for simplicity
Assuming the DB runs mostly at 0.25 CU (minimum) and is active 24×7:
* CU-hours per month =
* Compute cost =
* Storage cost =
Total estimated cost ≈ $47.15 / month
If the DB occasionally scales up to 0.5 CU on average, compute cost would roughly double to ~$93/month.
My questions:
1. If my DB mostly runs at 0.25 CU, will I be billed only for that usage, or is there a hidden minimum based on 1 CU?
2. How fast is autoscaling when Neon needs to scale from 0.25 CU → 1 CU under load? Any noticeable latency?
3. Based on my Aurora usage (2 vCPU / 16 GB RAM, ~7% CPU, 1 GB disk), does my math make sense, or am I missing something?
I’m trying to compare the $47–93/month Neon estimate versus my current Aurora cost (~$50–150/month for db.r7g.large) and decide if Neon is viable for production.
Thanks in advance for any guidance!
I’m currently running a production-grade app on AWS Aurora, and I’m considering switching to Neon Scale plan. I’d like to understand the pricing implications based on my real-world usage. Here are my current specs on Aurora:
* Instance:
db.r7g.large* CPU utilization: ~7% on average
* Disk usage: 1 GB (assume total size)
I want to migrate my main production database to Neon. Based on Neon Scale pricing:
* Compute is billed in CU-hours, where 1 CU = 1 vCPU + 4 GB RAM.
* Neon can autoscale down to 0.25 CU (0.25 vCPU + 1 GB RAM).
* My DB is mostly idle, so it would run near this minimum capacity most of the time.
* Neon Scale plan rate: $0.26 per CU-hour
* Storage: $0.35 per GB-month
* Egress / PITR and other fees are ignored for simplicity
Assuming the DB runs mostly at 0.25 CU (minimum) and is active 24×7:
* CU-hours per month =
0.25 CU × 24 h × 30 d ≈ 180 CU-hours* Compute cost =
180 × $0.26 ≈ $46.80 / month* Storage cost =
1 GB × $0.35 ≈ $0.35 / monthTotal estimated cost ≈ $47.15 / month
If the DB occasionally scales up to 0.5 CU on average, compute cost would roughly double to ~$93/month.
My questions:
1. If my DB mostly runs at 0.25 CU, will I be billed only for that usage, or is there a hidden minimum based on 1 CU?
2. How fast is autoscaling when Neon needs to scale from 0.25 CU → 1 CU under load? Any noticeable latency?
3. Based on my Aurora usage (2 vCPU / 16 GB RAM, ~7% CPU, 1 GB disk), does my math make sense, or am I missing something?
I’m trying to compare the $47–93/month Neon estimate versus my current Aurora cost (~$50–150/month for db.r7g.large) and decide if Neon is viable for production.
Thanks in advance for any guidance!
