Block storage types and how to use block storage - Elastic Compute Service

Cloud disks

Cloud disks use distributed storage to provide high data reliability for ECS instances.

Disk roles

Disk type	Description
System disk	Boot disk for an ECS instance. Stores the operating system and program files. Created with the instance and shares its lifecycle.
Data disk	Stores user data, logs, and applications. Can be created with an instance or separately.

Storage redundancy

Cloud disks support zone-redundant storage (ZRS) and locally redundant storage (LRS), spanning multiple zones or a single zone to ensure data durability.

Zone-redundant storage (ZRS)

ZRS stores data copies across multiple zones in the same region, with physically isolated IDCs, racks, and power supplies. This provides 99.9999999999% (12 nines) data reliability. If a zone fails, ZRS continues read and write services.

Disk type: Regional ESSDs
Use cases:
- Multi-zone disaster recovery for databases, big data, middleware, ERP/CRM.
- Container deployments across zones.
Important
ZRS writes data to multiple zones, so write latency may be higher than LRS. See Block storage performance.

If the entire region fails, data becomes inaccessible. For cross-region availability, create an automatic snapshot policy to regularly copy snapshots to other regions.

Locally redundant storage (LRS)

LRS stores data copies on multiple devices in different facilities within one zone, providing 99.9999999% (9 nines) data reliability and protecting against hardware failure.

Disk types: ESSD AutoPL disk, ESSD and ESSD Entry disk.
Use cases:
- ESSDs:
  - Online Transaction Processing (OLTP) databases (MySQL, PostgreSQL, Oracle), NoSQL databases (MongoDB, HBase, Cassandra) and Elasticsearch distributed logging.
  - System disk or replacement for Ultra/Basic disk.
- ESSD AutoPL disks:
  - Fixed capacity with variable performance needs.
  - Workloads with frequent traffic spikes (bursting support).
  - Replacement for Standard SSD.
- ESSD Entry:
  - Development and testing.
  - System disk or replacement for Ultra/Basic disk.
  ESSD Entry disks can be attached only to universal instance families (U instances) and the e, economy instance family.

Important

Data is stored in one zone. If the zone fails, data becomes inaccessible. For higher availability, use Regional ESSDs.

Example: Regional ESSDs vs ESSD PL1

Feature	Regional ESSDs	ESSD PL1
Redundancy	Zone-redundant storage (ZRS)	Locally redundant storage (LRS)
Durability	99.9999999999% (12 nines)	99.9999999% (9 nines)
Max IOPS (Input/Output Operations Per Second)	50,000	50,000
Max throughput (MB/s)	350	350
Avg write latency (4K blocks)	Millisecond-level	0.2 ms
Attachment scope	Attach to instances in any zone within the region.	Attach only to instances in the same zone as the disk.
Single-zone failure impact	No service disruption	Service unavailable
Price (Hangzhou, monthly)	0.2295 USD/GB/month	0.153 USD/GB/month

Performance tiers

Cloud disks include ESSD-series disks and previous-generation disks (standard SSDs, ultra disks, and basic disks).

Disk performance varies by type. Select a disk that meets your workload requirements.

For pricing and billing of different disk types, see Cloud disk billing.

ESSD-series disks

Disk type	Features	Scenarios	Data reliability guarantee	Billing
ESSDs	High IOPS Low latency	Latency-sensitive or I/O-intensive workloads: Large-scale OLTP databases NoSQL databases Elasticsearch distributed logging	99.9999999%	Capacity fee
ESSD AutoPL disks	Decouples capacity from performance. Supports provisioned performance. Adjust performance independently of capacity. Supports performance burst. Temporarily boosts disk performance to handle sudden I/O spikes.	ESSD-suitable scenarios Fixed capacity with high performance needs Workloads with frequent peaks requiring burst capacity	99.9999999%	Capacity fee Pay-as-you-go provisioned performance fee (charged after being enabled) Pay-as-you-go burst performance fee (charged after being enabled)
Regional ESSDs	High input/output operations per second (IOPS) Zone-redundant storage	ESSD-suitable scenarios Multi-zone disaster recovery for databases Cross-zone container deployment Self-built or cloud-deployed SaaS services	99.9999999999%	Capacity fee
ESSD Entry disk ESSD Entry disks can be attached only to universal instance families (U instances) and the e, economy instance family.	High IOPS Low latency	Development and testing System disk	99.9999999%	Capacity fee

Previous-generation disks

Standard SSDs, ultra disks, and basic disks are being phased out in some regions. Use PL0 ESSDs or ESSD Entry disks to replace ultra disks and basic disks, and ESSD AutoPL disks to replace standard SSDs.

Disk type	Features	Scenarios	Billing
Standard SSD	High random read and write performance High reliability	I/O-intensive applications Small and medium-sized relational databases and NoSQL databases	Capacity fee
Ultra disk	High cost-effectiveness High reliability	Development and testing workloads Use as a system disk	Capacity fee
Basic disk	High cost-effectiveness	Low-cost development and testing without high performance needs	Capacity fee

Local disks

Local disks reside on the physical machines hosting ECS instances and suit workloads requiring high I/O performance, mass storage, and cost-effectiveness.

Category	Supported instance family	Scenario
Local non-volatile memory express (NVMe) SSD	The following instance families use local NVMe SSDs: Instance families equipped with local SSDs: i4, i4g, i4r, i3, i3g, i2, i2g, i2ne, i2gne, and i1 GPU-accelerated compute-optimized instance family: gn5	Instance families equipped with local NVMe SSDs are suitable for the following scenarios: I/O-intensive applications that require high I/O performance and low latency, such as online gaming, e-commerce, live streaming, and media Applications that require high storage I/O performance and a high-availability architecture at the application layer, such as NoSQL databases (including Cassandra, MongoDB, and HBase), Massively Parallel Processing (MPP) data warehouses, and distributed file systems
Local SATA HDD	The d3s, d2c, d2s, d1ne, and d1 big data instance families use local SATA HDDs.	Local SATA HDDs are the preferred storage media for industries such as Internet and finance that have high requirements for big data computing, storage, and analytics. These disks are suited for mass storage and offline computing scenarios and can meet the high requirements of distributed computing services such as Hadoop in terms of storage performance, storage capacity, and internal network bandwidth.

Note

For performance details of local SSD-equipped and big data instance families, see Instance families.

Elastic ephemeral disks

An elastic ephemeral disk provides temporary, high-performance storage for ECS instances. It can be created with an instance or separately, and its capacity is customizable.

Block storage data security

Note

Except for data erasure, the following applies only to cloud disks, not to local disks or elastic ephemeral disks.

Read and write stability

Data is stored in triplicate across a block storage cluster in the same zone, ensuring read/write stability and 99.9999999% data reliability. See Triplicate technology for cloud disks.
Proactive backup

Periodically create snapshots to back up cloud disk data, including logs and customer transactions.
Data erasure mechanism

Deleted data is inaccessible to other users. When you delete data from the distributed block storage system, it is completely erased:
- Cloud disks use sequential append-writes at the underlying layer. This design leverages the high bandwidth and low latency of sequential writes on physical disks. Because of the append-write feature, an operation to delete a logical space on a cloud disk is recorded only as metadata. If you attempt to read from this logical space, the storage system returns all zeros. Similarly, overwriting a logical space does not immediately overwrite the corresponding space on the physical disk. Instead, the storage system modifies the mapping between the logical and physical spaces to perform the overwrite. This ensures that the original data cannot be read. Residual data on the physical disk from delete or overwrite operations is later permanently deleted.
- When you release a block device, such as a cloud disk, the storage system immediately destroys its metadata to make the data inaccessible. The physical storage space that the cloud disk occupied is also reclaimed. This physical space is cleared before it is reallocated. All newly created cloud disks return zeros for all read operations before the first write.
Data encryption

For sensitive data, encrypt cloud disks and their snapshots with the industry-standard AES-256 algorithm. Data is encrypted when written from an ECS instance to a disk and decrypted when read.

References

For block storage limits and quotas, see Limits of Block Storage.
For common cloud disk operations, see Guide to creating and using disks.
Alibaba Cloud also offers Object Storage Service (OSS) and File Storage NAS for other storage scenarios. See How do I select a storage service from NAS, OSS, and EBS?.