Overview

Block Storage is a redundant and distributed block device implementation for Exoscale Compute instances which provides rapid data access and retrieval to support your most demanding workloads with exceptional performance for latency-sensitive applications. Effortlessly meeting growing storage demands, and easily store containerized applications.

• High data durability
  Multiple data copies are maintained at all times on separate servers. Snapshots are retained indefinitely.
• Ultra-low latency for high-performance workloads
  Experience outstanding performance fueled by the latest generation NVMe SSDs—ideal for latency-sensitive applications such as databases.
• Flexible storage capabilities
  Add volumes to your instances anytime, enabling you to scale capacity as your needs evolve seamlessly.

Terminology

• Volume
  A single storage device that is partitioned and formatted to store directories and files.
• Snapshot
  A storage snapshot is the state of a volume at a particular point in time. It provides the ability to create new volumes with the state of the current volume at that specific point in time.

Features

Block Storage volumes feature a fixed 4KiB block size and replicate data at least twice within the same Zone for durability. They support snapshots—enabling the creation of new volumes (“forking”)—though exporting or downloading snapshots is impossible. Access is ReadWriteOnce, allowing only a single instance to attach a volume at any given time; however, volumes can be detached from one instance and reattached to another within the same Zone.

• IOPS
  Enjoy 5,000 IOPS per volume for read and write operations—no sharing involved.
• Built-In Feature
  Use Block Storage alongside Compute Instances or Managed Kubernetes.
• Easy Snapshots
  Take a snapshot of any volume via an API call. Create new volumes from snapshots whenever needed.
• Five Volumes Per Instance
  You can attach up to five Block Storage volumes to each instance, with a maximum capacity of 10 TiB per volume.
• Easy Scaling
  Create a Block Storage volume in seconds using the Exoscale Portal, CLI, API, or Terraform provider. Detach and reattach volumes to move them between Instances.
• Exoscale CSI Driver
  Interact seamlessly with Block Storage from Kubernetes using our Exoscale CSI Driver. (Read more)
• Two-Times Replication
  Data is replicated twice across multiple nodes, ensuring availability even if a disk fails.
• Secure Data Storage in Europe
  As a fully European cloud provider, all data remains in your chosen zone and is fully GDPR-compliant.

With its unique features, Block Storage suits mission-critical, I/O-intensive applications. Use it for relational and transactional databases, time series, container volumes, or hypervisor file systems. Because volumes are directly connected to an instance, their I/O speeds are often significantly faster than alternative storage solutions like Object Storage.

• Flexibility for Containerized Applications
  As containers are flexible, scalable, and efficient, so is our Block Storage. Containers benefit from Block Storage’s speed and the native ability to mount multiple volumes on a single host. Migrate seamlessly between servers, locations, and operating systems.
• Discover the Best Cost-to-Performance Ratio
  Our Block Storage offers a fair and transparent cost-to-performance ratio. With a simple pricing structure, you’ll have predictable monthly expenses—no cumbersome IOPS provisioning or pay-per-use variability.
• Persistent Volumes
  Create persistent volumes for critical information, databases, or other essential data to retain it over extended periods. These volumes can be accessed across multiple instances or sessions.

Availability

Limitations

The following table shows which limits are enforced on Block Storage usage. It also highlights some of the characteristics of the product:

In addition the default Organization Quotas are set and can be expanded upon request:

As block storage volumes expose a block size of 4KiB, some software that incorrectly assume a smaller block size might not work. This issue can be circumvented by using a compatibility layer in your instance like dm-ebs on linux.