OSS provides unlimited, secure, cost-effective, and reliable storage — a preferred choice for storing static resources and objects. Two features accelerate access to OSS: CDN acceleration and transfer acceleration. They use different mechanisms and are optimized for different workloads. CDN prioritizes high cache hit ratios and lower bandwidth costs for frequently accessed static content. Transfer acceleration prioritizes end-to-end latency reduction for uploads, dynamic data, and long-distance transfers.
How each method works
CDN acceleration
CDN is a distributed network built on top of a bearer network. It consists of points of presence (POPs) distributed across regions around the globe. Static data in OSS is cached on these POPs, so clients retrieve data directly from the nearest POP rather than from OSS. This reduces back-to-origin traffic, prevents network congestion, and ensures content is delivered with minimal latency.
Transfer acceleration
Transfer acceleration uses OSS data centers distributed around the globe. When a request reaches an acceleration endpoint, it is parsed and routed to the data center where the bucket is located over the optimal network path and protocol. This delivers an optimized end-to-end acceleration solution for access to OSS over the Internet.
When to use CDN acceleration
CDN acceleration is the better fit when:
Your workload serves frequently accessed, static resources — websites, applications, or media files that do not change often.
Reducing bandwidth costs is a priority, and high cache hit ratios are achievable for your content.
Data is expected to be downloaded many times and strong consistency is not required.
Access originates from a third-party source.
Typical use cases: websites or applications that host medium- and small-sized files, large file downloads, and on-demand audio and video streaming.
Limitation: CDN provides a high cache hit ratio for frequently accessed resources, but a low cache hit ratio for infrequently accessed ones. When a cache miss occurs, the request is redirected to OSS over the public network. CDN does not provide noticeable performance improvements for dynamic operations such as uploads and deletions.
When to use transfer acceleration
Transfer acceleration is the better fit when:
Data is uploaded or downloaded across continents or over long distances, particularly for gigabyte- to terabyte-scale objects.
Strong consistency is required — all requests go directly to OSS, so data is always up to date.
Your business stores dynamic resources that change frequently (CDN caching would serve stale content).
Data is infrequently accessed (cold data) and not worth caching on CDN POPs.
You need to accelerate both uploads and downloads, not just reads.
Limitation: Because all resource requests are made directly to OSS, transfer acceleration consumes a large amount of bandwidth. It also provides lower acceleration performance than CDN when a large number of requests for the same resource originate from the same region. Only HTTPS requests are supported.
Comparison
| CDN acceleration | Transfer acceleration | |
|---|---|---|
| How it works | Caches OSS data on globally distributed POPs; clients retrieve from the nearest POP | Routes requests to OSS over the optimal network path; requests still made to OSS |
| Best for | Frequently accessed static content; large file downloads; on-demand streaming | Long-distance transfers; GB–TB uploads and downloads; dynamic and cold data |
| Strengths | Large global POP network; high bandwidth capacity; reduces bandwidth costs | Accelerates uploads and downloads of large objects over long distances; supports all OSS features |
| Limitations | Low cache hit ratio for cold data; cache miss falls back to OSS over public network; limited benefit for uploads and deletions | Higher bandwidth consumption; lower performance than CDN for high-concurrency same-region requests; HTTPS only |