Object Storage Service:FAQ about OSS

What is Alibaba Cloud OSS?

Alibaba Cloud OSS is a secure, cost-effective, highly durable, and scalable storage service for large volumes of data. OSS is designed to provide data durability of at least 99.9999999999% (twelve 9's) and service availability of at least 99.995%.

What are the main features of OSS?

OSS supports RESTful API operations that work independently of the OSS console, so you can store and access data from any application, anywhere. Storage scales automatically with your usage — you pay only for what you use, with no performance or durability trade-offs.

Beyond the API, OSS provides SDKs and migration tools for transferring large volumes of data. OSS also offers multiple storage classes for different scenarios: store images, audio, and video as Standard objects for frequent access, and move infrequently accessed data to Infrequent Access (IA), Archive, Cold Archive, or Deep Cold Archive to reduce costs over time.

For the full feature set, see Functions and features.

Who uses OSS?

OSS is well-suited for anyone who needs to store large volumes of data. Common user groups include app and software developers, game development companies, and operators of online communities, media-sharing sites, and e-commerce platforms.

Common use cases:

• Audio, video, and image applications: Short video storage, live recording, video on demand, photo-sharing, and video albums.

• Education: Online education platforms (including K12) that need to store large content libraries and serve them via Alibaba Cloud CDN.

• AI and IoT: Autonomous driving teams can migrate training data to OSS using Data Transport. IoT video surveillance systems can upload camera feeds directly to OSS for real-time mobile viewing, with lifecycle rules handling tiered retention.

• Cinematic rendering: Large, scalable storage for filmmaking and media assets, with optional Intelligent Media Management (IMM) integration for intelligent processing.

• Genomics: Scalable storage for gene sequencing, delivery, and diagnostics data, with access to integrated cloud compute for analytics and machine learning.

What types of data does OSS store?

OSS is suited for attachments, high-definition images, audio and video objects, backup files for forums, websites, and software applications, and objects for file synchronization and online storage systems.

What are the advantages of OSS over local storage?

OSS lets developers tap into Alibaba Cloud's economy of scale without up-front investment or performance degradation. There are no capacity ceilings or performance bottlenecks tied to business growth, and no security infrastructure to manage yourself.

What is the maximum data volume OSS supports?

OSS does not impose limits on total storage capacity or on the storage capacity of any single bucket.

For individual object uploads:

• OSS console: Up to 5 GB per object.

• Larger than 5 GB: Use multipart upload, Graphical management tool ossbrowser 1.0, or ossutil 2.0.

What storage classes does OSS provide?

OSS offers five storage classes to cover scenarios from hot to cold data:

The core trade-off: as storage cost goes down (Standard to Deep Cold Archive), retrieval cost and latency go up.

For details, see Storage class.

How do I choose a storage class?

The right storage class depends on how often you access the data and your tolerance for retrieval latency.

As storage cost goes down (Standard to Deep Cold Archive), retrieval cost and latency go up. Archive, Cold Archive, and Deep Cold Archive data must be restored before access, which takes minutes to hours and incurs data retrieval fees.

As a starting point: if 70% of your data is not accessed for more than 30 days, move that data to IA or Archive. Configure lifecycle rules to automate these transitions so OSS handles the class changes based on object age.

Does Alibaba Cloud use data stored in OSS?

Alibaba Cloud does not use or disclose your data without your authorization. Data is processed only based on your service requirements or as required by applicable laws and regulations. For details, see the Alibaba Cloud International Website Product Terms of Service.

Does Alibaba Cloud use OSS internally?

Yes. Alibaba Cloud developers use OSS to store authorized data across various internal projects, relying on it for key business operations.

How does OSS handle traffic spikes?

OSS is designed to absorb traffic spikes generated by internet applications. Pay-as-you-go pricing and unlimited capacity keep your service running without interruption, and OSS load-balancing prevents spikes from affecting application performance.

How is data organized in OSS?

OSS stores data as key-value pairs. Each object is identified by a key (the object name), which you use to retrieve its content.

OSS uses a flat structure — all objects live in buckets with no true directory hierarchy. However, OSS supports directory-like organization through key prefixes. Setting a key to dir/example.jpg creates an apparent dir directory and places example.jpg within it. Deleting dir/example.jpg also removes the dir prefix.

What intelligent features does OSS provide?

OSS integrates with computing frameworks including Hadoop, Spark, MaxCompute, Batch Compute, High Performance Computing (HPC), and E-MapReduce (EMR). For media workflows, OSS works with IMM to handle storage and intelligent media processing. OSS also provides built-in SaaS services for image processing and content detection.

How do I get started with OSS?

1. Create an Alibaba Cloud account if you don't already have one.

2. Activate OSS from the activation page.

3. (Optional) The default billing method is pay-as-you-go. To reduce costs for predictable usage, purchase resource plans. See Purchase guide for resource plans.

4. Access OSS through the OSS console, ossbrowser, ossutil, or OSS SDKs for your preferred programming language. See Quick start.

What compliance certifications does OSS hold?

Alibaba Cloud holds major compliance certifications across Asia, Europe, and other countries and regions. Key certifications include:

• First globally: ISO 22301, CSA STAR gold certifications, and C5 additional requirements.

• First in Asia Pacific: C5 and ISO 27001 certified.

• First in China: MTCS Level 3 and ISO 20000 certified; ISO 27001 from BSI in 2012.

OSS also meets the compliance requirements of the Securities and Exchange Commission (SEC) and the Financial Industry Regulatory Authority (FINRA). Alibaba Cloud is the first cloud service provider in China to have passed the Cohasset Associates audit, following AWS, Azure, Google Cloud, and IBM.

For the full list, see Compliance certifications.

Why does accessing cold storage require data restoration?

Archive, Cold Archive, and Deep Cold Archive storage significantly reduces costs by trading off real-time access. Data stored in these classes is not immediately readable — it must be restored first, which takes minutes to hours. This model is intended for data you need to retain long-term but access rarely.

Where is my data stored?

When you create a bucket, you choose the Alibaba Cloud region where it lives. By default, OSS stores your data redundantly within a single zone in that region. Enable zone-redundant storage (ZRS) to store copies across multiple zones in the same region, so your data remains accessible even if one zone becomes unavailable.

What is an Alibaba Cloud region?

A region is a geographical area that contains multiple isolated zones connected by low-latency, high-throughput, and highly redundant networks.

What is a zone?

A zone is an isolated data center within a region. Zones in the same region are connected over the internal network. Network latency between instances in the same zone is lower than between instances in different zones. If one zone becomes unavailable, other zones in the same region are not affected.

How do I choose a region?

Consider physical proximity to your users, compatibility with other Alibaba Cloud services you use, and regional resource pricing. For guidance, see Choose an OSS region.

How does OSS billing work?

OSS uses pay-as-you-go billing — you pay only for what you use with no minimum fee. For predictable workloads, resource plans are often more cost-effective than pay-as-you-go. See the OSS pricing page for current pricing details.

Who pays when another account accesses my OSS resources?

By default, you (the bucket owner) are charged for all access to your resources, regardless of who initiates the request. Enable the pay-by-requester mode for your bucket so requesters are charged for the requests they send and the data they download. See Pay-by-requester.

How do I stop being charged for OSS?

OSS does not offer a deactivation feature, as deactivation would disrupt running workloads. To stop charges, delete your OSS resources instead. See How do I deactivate OSS or stop OSS charging my resources? for step-by-step guidance.

Is data in OSS secure?

By default, only the bucket owner can access resources. OSS provides identity verification and access control at both the bucket and object level. Access control lists (ACLs), RAM policies, and bucket policies let you grant specific permissions to specific users and groups.

The OSS console flags any buckets configured for public access. If a private bucket or object is reconfigured to public-read or public-read-write, OSS surfaces a warning. For a full overview, see Security and compliance overview.

How do I control access to my OSS data?

OSS provides multiple access control methods to allow users to access objects stored in buckets:

• ACLs: Set bucket-level or object-level permissions (private, public-read, or public-read-write).

• RAM policies: Grant fine-grained permissions to specific users and user groups.

• Bucket policies: Define resource-based policies attached directly to a bucket.

For a full comparison, see Overview of permissions and access control.

What encryption options does OSS provide?

OSS supports two encryption approaches:

• Server-side encryption: OSS automatically encrypts objects when uploaded and decrypts them when downloaded. The x-oss-server-side-encryption response header confirms server-side encryption is active. See Server-side encryption.

• Client-side encryption: Objects are encrypted on the client before upload. See Client-side encryption.

How do I protect objects from accidental deletion or overwrite?

Enable versioning on your bucket. When versioning is active, overwritten or deleted objects are preserved as previous versions rather than permanently removed. Restore any object to any previous version at any time. See Versioning.

What is a retention policy?

A retention policy enforces the Write Once Read Many (WORM) strategy that prevents an object from being deleted or overwritten for a specified period. After a retention policy is configured and locked for a bucket, you can read objects from or upload objects to the bucket. However, objects in the bucket or the retention policy cannot be deleted within the retention period specified in the retention policy.

Retention policies are useful for immutable records such as medical files, technical documents, and contracts.

Can I edit an object directly in OSS?

No. OSS does not support in-place object modification. To update an object, download it, make your changes, and re-upload it.

Does OSS use the triplicate storage mechanism?

No. OSS uses erasure coding (EC), which provides equivalent storage performance and reliability.

How is the 99.995% service availability calculated?

The OSS SLA defines availability using these formulas:

• Error rate per 5 minutes = Failed requests per 5 minutes / Total valid requests per 5 minutes × 100%

• Availability = (1 − Error rate per 5 minutes during the service period / Total number of errors in 5 minutes during the service period) × 100%

See the Alibaba Cloud International Website Object Storage Service Service Level Agreement for the full terms.

How do I replicate data to a bucket in a different region?

Use cross-region replication (CRR). CRR automatically and asynchronously replicates objects across buckets in different OSS regions in near real-time. Create, overwrite, and delete operations are all replicated from the source bucket to the destination. Configure multiple CRR rules on a single bucket for copies across multiple regions.

What are the use cases for CRR?

• Regulatory compliance: Store data copies at a physical distance to meet data residency regulations.

• Lower latency: Place object replicas closer to users in geographically distant locations.

• Disaster recovery: Maintain a live copy in a second region so your data remains accessible if one region is affected by a disaster.

• Data migration: Move data between regions to support business continuity.

• Multi-region compute: Allow compute clusters in two regions to analyze the same set of objects.

How is CRR billed?

CRR generates cross-region traffic charges for each replication operation, billed pay-as-you-go. Each synchronized object also counts as a request, billed at the standard request rate. Resource plans do not apply to CRR traffic — CRR is always charged pay-as-you-go.

How do I query data stored in OSS?

Use the SelectObject operation to run SQL statements against CSV or JSON objects without downloading the entire file. SelectObject filters data server-side, returning only the matching subset. It is well-suited for multipart queries on large objects, JSON object querying, and log analysis.

See Query objects for usage details.

What is lifecycle management and how does it reduce costs?

Lifecycle rules automatically transition objects to lower-cost storage classes or delete them after a specified period, based on object age. This eliminates manual management of cold data and reduces storage costs over time.

Example scenarios:

• A medical institution stores records that are accessed occasionally in the first six months after upload, then almost never. Configure a lifecycle rule to transition these objects to Archive 180 days after upload.

• A company stores call records that are frequently accessed for the first two months, occasionally from month two to six, and never after. Configure a rule to transition to IA at 60 days, Archive at 180 days, and delete at 730 days.

• A bucket contains a large number of objects you want to delete entirely. Because the console allows manual deletion of up to 1,000 objects at a time, configure a lifecycle rule to delete all objects one day later — all objects are cleared the next day.

For setup instructions, see Lifecycle rules based on the last modified time.

How do I get a periodic inventory of objects in a bucket?

Use the bucket inventory feature to export object information on a daily or weekly schedule. Each export includes object count, sizes, storage classes, and encryption status. See Bucket inventory.