What Is Block Storage? How It Works, Uses and Cloud Options

Quick answer: Block storage is a way of storing data in fixed size chunks called blocks, each with its own address, so a server can read or rewrite any single block directly without touching the rest. The operating system sees a block volume as a raw disk and puts its own file system on top. Because of this direct, low latency access, block storage runs databases, virtual machine disks and boot drives. In the cloud it appears as services such as Amazon EBS, Azure Managed Disks, Google Cloud Hyperdisk and OCI Block Volumes.

Key points

  • A block holds no file name, owner or folder path; the host’s file system tracks all of that, which keeps the storage layer fast and simple.
  • Networked block storage travels over Fibre Channel, iSCSI or NVMe over Fabrics, and the protocol largely sets latency and hardware cost.
  • Since September 2025, an Amazon EBS gp3 volume can reach 64 TiB, 80,000 IOPS and 2,000 MiB/s.
  • Cloud disks are usually less durable than object storage, so snapshots and backups still matter.
  • In warehousing, “block storage” means block stacking pallets on the floor, a completely different idea.

What is block storage?

Block storage is a data storage model that splits information into equal sized blocks and gives each one a numeric address. SNIA, the storage industry association, defines a logical block as data stored on a device and tied to an address for retrieval or overwriting.

Those addresses are called logical block addresses (LBAs). A block device presents itself as one long numbered row of blocks, usually 512 bytes or 4,096 bytes each. Nothing else lives at this layer: no file names, no permissions, no timestamps.

That is the key idea: the storage hands the server a blank disk, and the server decides how to organize it.

How does block storage work?

It works by letting the operating system format a volume with a file system such as ext4, XFS or NTFS, which then maps every file to a list of block addresses. When an app saves data, the file system translates the request into reads and writes against specific LBAs.

Here is the mechanism step by step:

  1. The server attaches a volume, either a local drive or a network volume presented as a logical unit number (LUN).
  2. The file system keeps an index of which blocks belong to which file.
  3. An application asks to change part of a file.
  4. The storage writes only the affected blocks, in place.

Step four explains why databases love this model. When PostgreSQL or Microsoft SQL Server updates one row, it rewrites a single small page on disk. Object storage cannot do that; changing an object means uploading a new version of the whole thing.

Networked volumes reach servers through a storage area network (SAN). Fibre Channel uses a dedicated fabric with its own switches and host bus adapters. iSCSI wraps SCSI commands in TCP/IP, so ordinary Ethernet works. NVMe over Fabrics carries native NVMe commands over RDMA, Fibre Channel or TCP for latency close to a local drive.

What is block storage used for?

It is used wherever software needs fast, random, small reads and writes: relational and NoSQL databases, virtual machine disks, operating system boot volumes and transactional applications like ERP systems.

Block storage examples you already use

  • The SSD inside your laptop is a block device, formatted with APFS or NTFS.
  • Every Amazon EC2 instance boots from an EBS volume or a local instance store, both block devices.
  • VMware VMFS datastores sit on SAN LUNs, which are block volumes.
  • Kubernetes PersistentVolumes often map to cloud disks through CSI drivers, such as the Persistent Disk driver on Google Kubernetes Engine.

In practice, when a team complains about “disk latency,” it is almost always talking about this layer.

Block storage vs file storage vs object storage

Block storage addresses raw blocks, file storage organizes data into folders shared over a network, and object storage keeps whole objects plus rich metadata behind an HTTP API. The right pick depends on how your application reads and writes.

FeatureBlock storageFile storageObject storage
Unit of dataFixed size blockFile in a folder treeObject with metadata
Who manages structureHost file systemStorage server (NAS)Storage service
Common protocolsFibre Channel, iSCSI, NVMe over FabricsNFS, SMBS3 compatible REST APIs
EditsRewrite single blocksRewrite filesReplace whole objects
Typical useDatabases, VMs, boot disksShared team foldersBackups, media, data lakes

What is block storage vs object storage?

Block storage wins on latency and in place edits; object storage wins on scale, cost per GB and durability. Amazon S3, for example, is designed for 11 nines of durability, while an EBS gp3 volume is designed for 99.8 to 99.9 percent. Our guide to the benefits of object storage covers that side in depth.

What is block storage vs file storage?

File storage, usually a NAS, lets many users open the same files at once over NFS or SMB. A block volume normally attaches to one server at a time, because two machines writing through ordinary file systems would corrupt each other’s data.

What is block storage in cloud computing?

Block storage in cloud computing is a network attached virtual disk that you create, size and attach to a virtual machine on demand. You pay for provisioned capacity and, on newer tiers, for the IOPS and throughput you dial in, rather than buying physical arrays.

CloudServiceNotable options (verified November 2027)
AWSAmazon EBSgp3 up to 64 TiB and 80,000 IOPS; io2 Block Express up to 256,000 IOPS and 4,000 MB/s
AzureManaged DisksFive types: Ultra Disk, Premium SSD v2, Premium SSD, Standard SSD, Standard HDD
Google CloudHyperdisk and Persistent DiskHyperdisk recommended; Local SSD is temporary storage
Oracle (OCI)Block VolumeOne volume type tuned by performance units, up to 300,000 IOPS per volume

What is block storage in AWS?

Block storage in AWS is Amazon Elastic Block Store (EBS). Its gp3 volumes give 3,000 IOPS and 125 MiB/s as a baseline at any size, with performance provisioned separately from capacity. For the heaviest databases, io2 Block Express adds durability of 99.999 percent and Multi Attach to as many as 16 Nitro based instances in one Availability Zone.

What is block storage in Azure?

Azure calls it Managed Disks. Ultra Disk and Premium SSD v2 let you set capacity, IOPS and throughput independently, but neither can serve as an OS disk, so the boot drive usually sits on Premium SSD or Standard SSD.

What is block storage in GCP?

Google Cloud offers durable block storage through Hyperdisk and Persistent Disk, plus Local SSD for scratch data that disappears when the VM stops. Google recommends Hyperdisk, and Persistent Disk is not available on its latest machine series.

What is block storage in OCI?

OCI Block Volume uses a single volume type and a performance slider measured in volume performance units (VPUs) per GB. Balanced uses 10 VPUs, Higher Performance 20, and Ultra High Performance 30 to 120. Volumes range from 50 GB to 32 TB and attach over iSCSI or paravirtualized connections.

Limitations and trade offs of block storage

Fast does not mean effortless. You pay for every provisioned gigabyte whether you fill it or not, volumes are usually tied to one zone, and sharing a volume between servers needs a cluster aware file system.

What most articles miss is durability. AWS designs gp3 for 99.8 to 99.9 percent annual durability, and OCI designs its volumes for 99.99 percent. Those are strong numbers, yet far below object storage. Most people discover this after a failed volume: snapshots to object storage, plus a solid 3 2 1 backup rule, are what actually protect the data. Set snapshot frequency from your RPO and RTO targets rather than guessing.

What is block storage in a warehouse?

In a warehouse, block storage usually refers to block stacking: placing loaded pallets directly on the floor and stacking them into dense blocks without racks. It has nothing to do with computer data.

Block stacking has a low setup cost because there is no racking to buy. However, pallets come out last in, first out (LIFO), it needs a lot of floor space, and it fails for fragile goods that cannot carry weight on top.

Frequently asked questions

Is a hard drive block storage?

Yes. Hard drives and SSDs expose numbered blocks to the computer, and the operating system’s file system decides which blocks hold which files.

Is block storage faster than object storage?

Usually, for small random reads and writes. Block volumes skip HTTP overhead and allow in place updates, so they suit databases. Object storage is built for large files and massive parallel access instead.

What is a LUN?

A LUN, or logical unit number, identifies a block volume that a SAN presents to a server. The server treats each LUN as its own disk.

Can multiple servers share one block volume?

Sometimes. AWS io2 Block Express supports Multi Attach, and some Google Cloud Hyperdisk types are shareable. The application or file system must coordinate writes to avoid corruption.

What is IOPS?

IOPS means input/output operations per second, the number of reads and writes a volume can handle each second. It matters most for databases doing many small operations.

Is block storage expensive?

It typically costs more per GB than object storage, and cloud providers bill provisioned capacity rather than used space, so rightsizing oversized volumes cuts waste directly.

What is thin provisioning?

Thin provisioning lets a storage system present a large volume while consuming physical space only as data is actually written. It saves capacity, but the pool must be monitored so it never fills up unexpectedly.

What is temporary or ephemeral block storage?

It is a drive physically attached to the host, such as Google Cloud Local SSD or an AWS instance store. It is very fast, but data is lost when the instance stops or fails.

Does Kubernetes use block storage?

Yes. Kubernetes PersistentVolumes connect to cloud disks through Container Storage Interface (CSI) drivers, giving stateful pods such as databases durable storage that survives restarts.

What is the difference between SAN and NAS?

A SAN delivers block storage over a dedicated storage network. A NAS delivers file storage over regular Ethernet using NFS or SMB.

Can block storage be encrypted?

Yes. All major clouds support encryption at rest for disks. Google Cloud encrypts Persistent Disk data automatically, and AWS lets you turn on EBS encryption by default for an account. On premises arrays typically offer self encrypting drives.

What file systems run on block storage?

Common choices are ext4 and XFS on Linux, NTFS and ReFS on Windows, APFS on macOS, and VMFS for VMware datastores.

Conclusion

Pick block storage when your application edits small pieces of data fast and often, and pair it with object storage for backups and bulk files. For exact limits, check the official Amazon EBS volume documentation, Microsoft’s Azure managed disk types guide and the SNIA storage dictionary for the underlying terms.

Leave a Reply

Your email address will not be published. Required fields are marked *