Instance Types
Nirvana Cloud offers predefined instance types organized into families optimized for different workloads.
You can also fetch the latest instance types programmatically via the API or the Terraform data source.
N1 Series
Section titled “N1 Series”The N1 series runs on AMD EPYC 7513 processors (32 Core / 64 Threads, 3.25 - 3.8 GHz) and is available in three families:
| Family | Prefix | vCPU : Memory Ratio | Best For |
|---|---|---|---|
| High CPU | n1-highcpu | 1:2 | Compute-intensive workloads such as batch processing, CI/CD, and high-traffic web servers |
| Standard | n1-standard | 1:4 | General-purpose workloads with a balanced mix of compute and memory |
| High Memory | n1-highmem | 1:8 | Memory-intensive workloads such as databases, caching layers, and data indexing |
Naming Convention
Section titled “Naming Convention”Instance type names follow the pattern: {generation}-{family}-{vCPUs}
For example, n1-standard-8 is a first-generation standard instance with 8 vCPUs and 32 GB of memory.
High CPU (n1-highcpu)
Section titled “High CPU (n1-highcpu)”Optimized for compute-intensive workloads that require high vCPU performance relative to memory. Each vCPU is paired with 2 GB of memory.
| Instance Type | vCPUs | Memory (GB) | Region Availability |
|---|---|---|---|
n1-highcpu-2 | 2 | 4 | us-sva-2 |
n1-highcpu-4 | 4 | 8 | us-sva-2 |
n1-highcpu-8 | 8 | 16 | us-sva-2 |
n1-highcpu-16 | 16 | 32 | us-sva-2 |
n1-highcpu-32 | 32 | 64 | us-sva-2 |
n1-highcpu-48 | 48 | 96 | us-sva-2 |
n1-highcpu-64 | 64 | 128 | us-sva-2 |
n1-highcpu-96 | 96 | 192 | us-sva-2 |
Standard (n1-standard)
Section titled “Standard (n1-standard)”Balanced instances for general-purpose workloads. Each vCPU is paired with 4 GB of memory.
| Instance Type | vCPUs | Memory (GB) | Region Availability |
|---|---|---|---|
n1-standard-2 | 2 | 8 | us-sva-2 |
n1-standard-4 | 4 | 16 | us-sva-2 |
n1-standard-8 | 8 | 32 | us-sva-2 |
n1-standard-16 | 16 | 64 | us-sva-2 |
n1-standard-32 | 32 | 128 | us-sva-2 |
n1-standard-48 | 48 | 192 | us-sva-2 |
n1-standard-64 | 64 | 256 | us-sva-2 |
n1-standard-96 | 96 | 384 | us-sva-2 |
n1-standard-128 | 128 | 512 | us-sva-2 |
n1-standard-192 | 192 | 768 | us-sva-2 |
High Memory (n1-highmem)
Section titled “High Memory (n1-highmem)”Optimized for memory-intensive workloads that require large amounts of RAM relative to vCPUs. Each vCPU is paired with 8 GB of memory.
| Instance Type | vCPUs | Memory (GB) | Region Availability |
|---|---|---|---|
n1-highmem-2 | 2 | 16 | us-sva-2 |
n1-highmem-4 | 4 | 32 | us-sva-2 |
n1-highmem-8 | 8 | 64 | us-sva-2 |
n1-highmem-16 | 16 | 128 | us-sva-2 |
n1-highmem-32 | 32 | 256 | us-sva-2 |
n1-highmem-48 | 48 | 384 | us-sva-2 |
n1-highmem-64 | 64 | 512 | us-sva-2 |
n1-highmem-96 | 96 | 768 | us-sva-2 |
What are instance types?
Section titled “What are instance types?”Instance types are predefined configurations of vCPU and memory, each optimized for a specific workload profile. Instead of configuring resources manually, you pick a family that matches your workload, choose a size, and deploy. Standardized types mean predictable pricing, easy comparison across clouds, and simpler capacity planning.
What are the three families and when should I use each?
Section titled “What are the three families and when should I use each?”n1-standard(4 GB per vCPU) — general-purpose. Web servers, application backends, relational databases, API services, microservice fleets. This is where most teams start.n1-highmem(8 GB per vCPU) — memory-optimized. In-memory caches (Redis, Memcached), analytical databases (ClickHouse, DuckDB), blockchain archive nodes, data warehousing. Use this when your workload needs large memory-to-CPU ratios.n1-highcpu(2 GB per vCPU) — compute-bound work. CI/CD pipelines, video encoding, cryptographic verification, blockchain validation. Use this when CPU cycles matter more than memory.
How does the naming convention work?
Section titled “How does the naming convention work?”Format: n1-{family}-{vcpus}. The n1 prefix denotes the series and generation. The family describes the memory-to-CPU ratio. The number is the vCPU count. For example, n1-standard-16 is a first-generation standard instance with 16 vCPUs and 64 GB of memory (4 GB per vCPU).
How do Nirvana instance types map to GCP and AWS?
Section titled “How do Nirvana instance types map to GCP and AWS?”| Nirvana | GCP Equivalent | AWS Equivalent | Memory Ratio |
|---|---|---|---|
n1-standard | n2-standard | m6a | 4 GB/vCPU |
n1-highmem | n2-highmem | r6a | 8 GB/vCPU |
n1-highcpu | n2-highcpu | c6a | 2 GB/vCPU |
If you’re coming from either platform, pick the same family name and the same vCPU count — the workload behaves the same way.
Do instance types work with NKS?
Section titled “Do instance types work with NKS?”Yes. NKS node pools support instance type selection. Specify a type like n1-standard-4 in your node pool config, and the scheduler respects it.
Can I still request custom configurations?
Section titled “Can I still request custom configurations?”Custom configs are available on a case-by-case basis. For the vast majority of workloads, one of the 26 standard types fits. If you need something specific, talk to sales — but start with standard types first.
What does “n1” mean? Will there be n2?
Section titled “What does “n1” mean? Will there be n2?”n1 is the first-generation instance family running on AMD EPYC 7513 processors. When new chipsets ship (newer CPUs, faster memory, better networking), the generation will increment to n2.