Compute Add-ons
Every project on the Supabase Platform comes with its own dedicated Postgres instance running inside a virtual machine (VM). The following table describes the base instance with additional compute add-ons available if you need extra performance when scaling up Supabase.
| Plan | Pricing | CPU | Memory | Connections: Direct | Connections: Pooler |
|---|---|---|---|---|---|
| Free (Included) | $0 | 2-core ARM (shared) | 1 GB | 60 | 200 |
| Small | $5 | 2-core ARM (shared) | 2 GB | 90 | 200 |
| Medium | $50 | 2-core ARM (shared) | 4 GB | 120 | 200 |
| Large | $100 | 2-core ARM (dedicated) | 8 GB | 160 | 300 |
| XL | $200 | 4-core ARM (dedicated) | 16 GB | 240 | 700 |
| 2XL | $400 | 8-core ARM (dedicated) | 32 GB | 380 | 1500 |
| 4XL | $950 | 16-core ARM (dedicated) | 64 GB | 480 | 3000 |
| 8XL | $1,860 | 32-core ARM (dedicated) | 128 GB | 490 | 6000 |
| 12XL | $2,790 | 48-core ARM (dedicated) | 192 GB | 500 | 9000 |
| 16XL | $3,720 | 64-core ARM (dedicated) | 256 GB | 500 | 12,000 |
Number of connections above are recommended values.
Contact us if you require a custom plan.
Dedicated vs. shared CPU#
All Postgres instances on Supabase are dedicated applications running inside dedicated virtual machines. However, the underlying hardware resources, for example the physical CPU, may be shared between multiple VMs, but appear to the OS as if it is a dedicated hardware CPU. This is commonly referred to as a vCPU (virtual CPU). Cloud providers use these shared hardware resources to save cost—you can upgrade to a larger compute add-on to guarantee a dedicated physical CPU for your instance.
Compute upgrades #
When considering compute upgrades, assess whether your bottlenecks are hardware-constrained or software-constrained. For example, you may want to look into optimizing the number of connections or examining query performance. When you're happy with your Postgres instance's performance, then you can focus on additional compute resources. For example, you can load test your application in staging to understand your compute requirements. You can also start out on a smaller tier, create a report in the Dashboard to monitor your CPU utilization, and upgrade later as needed
Disk IO#
SSD Disks are attached to your servers and the disk performance depends on the compute add-on of your instance. Disk IO refers to two metrics: throughput (Megabits per Second) and IOPS (Input/Output Operations per Second).
| Plan | Pricing | Max Disk Throughput | Baseline Disk Throughput | Max IOPS | Baseline IOPS |
|---|---|---|---|---|---|
| Free (Included) | $0 | 2,085 Mbps | 87 Mbps | 11,800 IOPS | 500 IOPS |
| Small | $5 | 2,085 Mbps | 174 Mbps | 11,800 IOPS | 1,000 IOPS |
| Medium | $50 | 2,085 Mbps | 347 Mbps | 11,800 IOPS | 2,000 IOPS |
| Large | $100 | 4,750 Mbps | 630 Mbps | 20,000 IOPS | 3,600 IOPS |
| XL | $200 | 4,750 Mbps | 1,188 Mbps | 20,000 IOPS | 6,000 IOPS |
| 2XL | $400 | 4,750 Mbps | 2,375 Mbps | 20,000 IOPS | 12,000 IOPS |
| 4XL | $950 | 4,750 Mbps | 4,750 Mbps | 20,000 IOPS | 20,000 IOPS |
| 8XL | $1,860 | 9,500 Mbps | 9,500 Mbps | 40,000 IOPS | 40,000 IOPS |
| 12XL | $2,790 | 14,250 Mbps | 14,250 Mbps | 50,000 IOPS | 50,000 IOPS |
| 16XL | $3,720 | 19,000 Mbps | 19,000 Mbps | 80,000 IOPS | 80,000 IOPS |
Contact us if you require a custom plan.
Bursting and Disk Budget#
Smaller compute instances can burst up to their largest throughput and IOPS for 30 minutes in a day. Beyond that, the performance reverts to the baseline. For example, the free tier can burst up to 2,085 Mbps for 30 minutes a day and reverts to the baseline performance of 87 Mbps. Your disk budget gets replenished throughout the day.
If you need consistent disk performance, choose the 4XL or larger compute add-on which has the same baseline and maximum disk throughput and IOPS.
If you're unsure of how much throughput or IOPS your application requires, you can load test your project and inspect these metrics in the Dashboard. If the Disk IO % consumed stat is more than 1%, it indicates that your workload has burst beyond the baseline IO throughput during the day. If this metric goes to 100%, the workload has used up all available disk budget and will revert to baseline performance. Projects that use any disk budget are good candidates for upgrading to a larger compute add-on with higher baseline throughput.