> ## Documentation Index
> Fetch the complete documentation index at: https://gcore.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Volumes

A volume is a block storage device that works like a physical disk but resides in the cloud. Volumes boot Virtual Machines by storing the operating system, and provide general-purpose storage for non-OS data.

We use a software-defined distributed file system to store volume data. It runs across a cluster of scalable nodes with triple data replication and load distribution, ensuring high availability.

## Boot and regular volumes

Each Virtual Machine requires a boot volume to start. The boot volume is created from an OS image or a custom image and appears as **System** on the VM creation form in the Customer Portal.

The boot volume cannot be deleted or detached while the Virtual Machine exists. When terminating a Virtual Machine, the boot volume can be kept independently — it is not deleted automatically — and used to create another Virtual Machine.

A regular volume is a general-purpose storage volume for any non-OS data.

<Info>
  Both boot and regular volumes support [snapshots](/cloud/virtual-instances/snapshots/take-a-snapshot-of-your-file-system).
</Info>

## Available volume types

The [Gcore Customer Portal](https://portal.gcore.com) supports the volume types listed below. Not all types are available in every region — the [Gcore Cloud calculator](https://gcore.com/pricing/cloud) shows availability by region. Configure each type while [creating a volume](/cloud/virtual-instances/volumes/create-and-configure-volumes).

| Volume             | Description                                                                                                                                                             |
| ------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| High IOPS SSD      | High-performance SSD block storage for latency-sensitive transactional workloads (60 IOPS per 1 GB; 2.5 MB/s per 1 GiB). IOPS limit: 9,000. Bandwidth limit: 500 MB/s.  |
| Standard           | Network SSD disk with stable high random I/O performance and high data reliability (6 IOPS per 1 GB; 0.4 MB/s per 1 GiB). IOPS limit: 4,500. Bandwidth limit: 300 MB/s. |
| Cold (deprecated)  | Network HDD disk for less frequently accessed workloads. Max IOPS: 1,000. Bandwidth limit: 100 MB/s.                                                                    |
| Ultra (deprecated) | Network block storage for non-critical, infrequently accessed data. Max IOPS: 1,000. Bandwidth limit: 100 MB/s.                                                         |
| SSD Low-Latency    | SSD block storage for applications requiring low-latency storage and real-time data processing. IOPS up to 5,000; average latency 300 µs.                               |

Snapshots and volume resizing are not supported for SSD Low-Latency volumes.

For High IOPS SSD and Standard volumes:

* Both volume IO burst and throughput burst are supported.
* IO and throughput credits exceed base limits and ensure optimal performance at peak IOPS and throughput.
* Credits increase IOPS and volume throughput by 10 times for 10 minutes.

For example, a Virtual Machine with a 3 GB High IOPS SSD volume has base limits of 180 IOPS (60 × 3) and 7.5 MB/s (2.5 × 3). When those limits are exceeded, burst credits apply:

* 180 IOPS × 10 = 1,800 IOPS
* 7.5 MB/s × 10 = 75 MB/s

Credits are renewed after peak usage subsides. When base limits are exceeded again, the renewed credits apply.