Blob 存储的可伸缩性和性能目标Scalability and performance targets for Blob storage

本参考详细介绍了 Azure 存储的可伸缩性和性能目标。This reference details scalability and performance targets for Azure Storage. 以下所列的可伸缩性和性能目标为高端目标,但却是能够实现的。The scalability and performance targets listed here are high-end targets, but are achievable. 在任何情况下,存储帐户实现的请求速率和带宽取决于存储对象大小、使用的访问模式、应用程序执行的工作负荷类型。In all cases, the request rate and bandwidth achieved by your storage account depends upon the size of objects stored, the access patterns utilized, and the type of workload your application performs.

请务必测试服务,确定其性能是否达到要求。Make sure to test your service to determine whether its performance meets your requirements. 如果可能,应避免流量速率突发峰值,并确保流量在各个分区上均匀分布。If possible, avoid sudden spikes in the rate of traffic and ensure that traffic is well-distributed across partitions.

当应用程序达到分区能够处理的工作负荷极限时,Azure 存储将开始返回错误代码 503(服务器忙)或错误代码 500(操作超时)响应。When your application reaches the limit of what a partition can handle for your workload, Azure Storage begins to return error code 503 (Server Busy) or error code 500 (Operation Timeout) responses. 如果发生 503 错误,请考虑修改应用程序以使用指数退避策略进行重试。If 503 errors are occurring, consider modifying your application to use an exponential backoff policy for retries. 使用指数退让策略,可以减少分区上的负载,缓解该分区的流量高峰。The exponential backoff allows the load on the partition to decrease, and to ease out spikes in traffic to that partition.

Blob 存储的缩放目标Scale targets for Blob storage

资源Resource 目标Target
单个 Blob 容器的最大大小Maximum size of single blob container 与最大存储帐户容量相同Same as maximum storage account capacity
块 Blob 或附加 Blob 中的块数上限Maximum number of blocks in a block blob or append blob 50,000 块50,000 blocks
块 Blob 中块的最大大小Maximum size of a block in a block blob 4000 MiB4000 MiB
块 Blob 的最大大小Maximum size of a block blob 50,000 X 4000 MiB(大约 190.7 TiB)50,000 X 4000 MiB (approximately 190.7 TiB)
附加 Blob 中块的最大大小Maximum size of a block in an append blob 4 MiB4 MiB
附加 Blob 的最大大小Maximum size of an append blob 50,000 x 4 MiB(大约 195 GiB)50,000 x 4 MiB (approximately 195 GiB)
页 Blob 的最大大小Maximum size of a page blob 8 TiB28 TiB2
每个 Blob 容器存储的访问策略的最大数目Maximum number of stored access policies per blob container 55
单个 blob 的目标请求速率Target request rate for a single blob 每秒最多 500 个请求Up to 500 requests per second
单个页 blob 的目标吞吐量Target throughput for a single page blob 最高每秒 60 MiB2Up to 60 MiB per second2
单个块 blob 的目标吞吐量Target throughput for a single block blob 上限为存储帐户的传入/传出限制1Up to storage account ingress/egress limits1

1 单个 blob 的吞吐量取决于多个因素,包括但不限于:并发性、请求大小、性能层、源上传速度和目标下载速度。1 Throughput for a single blob depends on several factors, including, but not limited to: concurrency, request size, performance tier, speed of source for uploads, and destination for downloads. 具体地说,对于标准存储帐户,请使用大于 4 MiB 的 Blob 或块大小调用 Put BlobPut Block 操作。Specifically, call the Put Blob or Put Block operation with a blob or block size that is greater than 4 MiB for standard storage accounts. 对于高级块 blob 或 Data Lake Storage Gen2 存储帐户,请使用大于 256 KiB 的块或 blob 大小。For premium block blob or for Data Lake Storage Gen2 storage accounts, use a block or blob size that is greater than 256 KiB.

2 在拥有“分层命名空间”设置的帐户中,尚不支持页 blob。2 Page blobs are not yet supported in accounts that have the Hierarchical namespace setting on them.

下表描述了服务版本允许的最大块大小和 blob 大小。The following table describes the maximum block and blob sizes permitted by service version.

服务版本Service version 最大块大小(通过放置块)Maximum block size (via Put Block) 最大 blob 大小(通过放置块列表)Maximum blob size (via Put Block List) 通过单个写入操作的最大 blob 大小(通过放置 Blob)Maximum blob size via single write operation (via Put Blob)
版本 2019-12-12 和更高版本Version 2019-12-12 and later 4000 MiB4000 MiB 大约 190.7 TiB(4000 MiB X 50,000 块)Approximately 190.7 TiB (4000 MiB X 50,000 blocks) 5000 MiB(预览版)5000 MiB (preview)
版本 2016-05-31 到版本 2019-07-07Version 2016-05-31 through version 2019-07-07 100 MiB100 MiB 大约 4.75 TiB(100 MiB X 50,000 块)Approximately 4.75 TiB (100 MiB X 50,000 blocks) 256 MiB256 MiB
2016-05-31 之前的版本Versions prior to 2016-05-31 4 MiB4 MiB 大约 195 GiB(4 MiB X 50,000 块)Approximately 195 GiB (4 MiB X 50,000 blocks) 64 MiB64 MiB

另请参阅See also