Accelerated Block Storage (ABS)

Accelerated Block Storage (ABS) is a high-performance, crypto-tuned block storage layer built for always-hot blockchain data with sustained IOPS, delivering high throughput, high availability, and cost-effective storage for I/O-intensive workloads.

ABS combines cloud elasticity with bare-metal performance, staying fast under sustained load without throttling, burst credits, or surprise bills.

Overview

Traditional cloud storage was designed for spiky, bursty workloads. Blockchain workloads are different: blocks keep coming, generating continuous I/O for hours or days. Indexers, archives, and trace-heavy nodes stay hot under constant load with no recovery window.

ABS is engineered to stay fast under continuous load. It delivers sustained IOPS indefinitely without time-based throttling, performance cliffs, or gradual degradation.

Availability

ABS is currently available in the us-sva-2 (Silicon Valley) region and being rolled out to all new regions.

Performance Specifications

IOPS	20K baseline (guaranteed) with up to 600K burst capacity
Latency	Sub-millisecond (~0.3 to 1.5 ms)
Throughput	Up to 1.6 GB/s writes, 260-380 MB/s sustained, ~600 MB/s reads
Volume Size	32 GB to 4 PB
Queue Depth	Near zero under sustained load

Key Features

Sustained Performance	No throttling after 2-3 hours like traditional cloud. Stays fast as long as your workload stays hot.
High Availability	Persistent shared volumes decoupled from compute. Data is never tied to a specific server. If a machine fails, workloads migrate instantly without rebuilds or data loss.
Elastic Scale	Scale from 32 GB to 4 PB with zero downtime. No hardware changes or re-provisioning required.
Predictable Pricing	Flat, TB-based pricing ($93.5/TB) with IOPS and throughput included. No burst credits, no metering.
Node-Level Colocation	Tightest physical data path when paired with Nirvana bare-metal compute for ultra-low latency.

Cost Comparison

ABS delivers significant cost savings compared to traditional cloud storage:

• 20x faster than gp3 and 60%+ cheaper
• 1.5x faster than io2 and 80%+ cheaper

Use Cases

ABS is purpose-built for workloads that require sustained I/O performance:

• RPC Nodes: Archive and trace-heavy nodes that stay hot under constant load
• Indexers: Continuous indexing workloads with heavy write patterns
• Databases: ClickHouse, Postgres, and OpenSearch clusters requiring consistent performance
• MPC Engines: Multi-party computation workloads with deterministic latency requirements
• Rollups: High-throughput rollup infrastructure
• Observability & Analytics: Data-intensive pipelines and ETL workloads
• Multi-Chain Data Platforms: Large archives and multi-chain datasets

Getting Started

ABS is available today for teams running I/O-intensive workloads. Contact us to start a POC.

Known Issues

Expanding Volumes

Currently, after creating an ABS Volume (via VM creation or independently) it cannot be expanded automatically. The Volume will expand but it won’t be registered in the OS correctly and thus the additional space will not be immediately usable.

As a temporary workaround, please reach out to the Nirvana team in Slack and we will carry out the Volume expansion.

FAQ

What is Nirvana ABS?

ABS is Accelerated Block Storage. Our high-ops, low-latency block storage solution available in our generation 2 data center region us-sva-2.

Does `us-sva-2` support only ABS?

Yes - us-sva-2 only supports our new Nirvana ABS Block Storage. We may in future introduce more options but for now only ABS is supported for both boot and data volumes.

Which regions is ABS available in?

ABS will be rolled out and available in all next-generation sites in the coming months.

What is the maximum size of a volume for Nirvana ABS?

ABS can support volumes up to 4 PB. Today customers can provision up to 100 TB automatically and should reach out to the Nirvana team if they want to provision larger volumes.

How do I use Nirvana ABS with the Terraform Provider?

Using the Nirvana Terraform Provider, set the volume type to abs.

For separately managed nirvana_compute_volume resources:

resource "nirvana_compute_volume" "example_compute_volume" {
  name = "my-data-volume"
  type = "abs"
  size = 100
  vm_id = nirvana_compute_vm.id
}

For nirvana_compute_vm with nested boot_volume and data_volumes fields:

resource "nirvana_compute_vm" "vm" {
  region = var.region

  name          = "my-vm"
  os_image_name = data.nirvana_compute_vm_os_images.vm_os_images.items[0].name
  ssh_key = {
    public_key = var.ssh_public_key
  }
  cpu_config = {
    vcpu = var.cpu
  }
  memory_config = {
    size = var.memory
  }
  boot_volume = {
    size = var.boot_disk_size
    type = "abs"
  }
  data_volumes = [
    {
      name = "my-data-volume"
      size = var.data_disk_size
      type = "abs"
    }
  ]

  subnet_id         = nirvana_networking_vpc.vpc.subnet.id
  public_ip_enabled = true

  tags = var.tags
}

What filesystems work best with Nirvana ABS?

All filesystems work with ABS but we recommend avoiding those with compression as this is handled at hardware level and may introduce additional latencies.