块 Blob 存储的性能层Performance tiers for block blob storage

由于企业部署的是性能敏感的云本机应用程序,因此在不同的性能级别上提供经济高效的数据存储选项非常重要。As enterprises deploy performance sensitive cloud-native applications, it's important to have options for cost-effective data storage at different performance levels.

Azure 块 Blob 存储提供两个不同的性能层:Azure block blob storage offers two different performance tiers:

  • 高级:已针对高事务率和 10 毫秒以下的稳定存储延迟进行优化Premium: optimized for high transaction rates and single-digit consistent storage latency
  • 标准:已针对高容量和高吞吐量进行优化Standard: optimized for high capacity and high throughput

以下注意事项适用于不同的性能层:The following considerations apply to the different performance tiers:

区域Area 标准性能Standard performance 高级性能层Premium performance
上市区域Region availability 所有区域All regions 特定区域In select regions
支持的存储帐户类型Supported storage account types 常规用途 v2、Blob 存储、常规用途 v1General purpose v2, BlobStorage, General purpose v1 BlockBlobStorageBlockBlobStorage
支持高吞吐量块 BlobSupports high throughput block blobs 是,在大于 4 MiB PutBlock 或 PutBlob 大小时Yes, at greater than 4 MiB PutBlock or PutBlob sizes 是,在大于 256 KiB PutBlock 或 PutBlob 大小时Yes, at greater than 256 KiB PutBlock or PutBlob sizes
冗余Redundancy 请参阅存储帐户的类型See Types of storage accounts 目前仅支持本地冗余存储 (LRS)Currently supports only locally-redundant storage (LRS)

在成本方面,高级性能层为具有高事务率的应用程序提供优化的定价,以帮助降低这些工作负荷的总存储成本Regarding cost, premium performance provides optimized pricing for applications with high transaction rates to help lower total storage cost for these workloads.

高级性能层Premium performance

高级性能层块 Blob 存储通过高性能硬件来提供数据。Premium performance block blob storage makes data available via high-performance hardware. 数据存储在已针对低延迟进行优化的固态硬盘 (SSD) 上。Data is stored on solid-state drives (SSDs) which are optimized for low latency. 与传统的机械硬盘相比,SSD 提供更高的吞吐量。SSDs provide higher throughput compared to traditional hard drives.

对于需要快速且一致的响应时间的工作负荷而言,高级性能存储是理想之选。Premium performance storage is ideal for workloads that require fast and consistent response times. 它最适合用于执行许多小型事务的工作负荷。It's best for workloads that perform many small transactions. 示例工作负荷包括:Example workloads include:

  • 交互式工作负荷Interactive workloads. 这些工作负荷需要即时更新和用户反馈,例如电子商务和地图应用程序。These workloads require instant updates and user feedback, such as e-commerce and mapping applications. 例如,在电子商务应用程序中,不经常查看的商品可能不会缓存。For example, in an e-commerce application, less frequently viewed items are likely not cached. 但是,这些商品必须可按需立即向客户显示。However, they must be instantly displayed to the customer on demand.

  • 分析Analytics. 在 IoT 方案中,每秒可能就会将大量的小规模写入操作推送到云中。In an IoT scenario, lots of smaller write operations might be pushed to the cloud every second. 可能会提取大量的数据,将其聚合进行分析,然后立即将其删除。Large amounts of data might be taken in, aggregated for analysis purposes, and then deleted almost immediately. 高级块 Blob 存储的大规模引入功能使之能够为此类工作负荷带来效益。The high ingestion capabilities of premium block blob storage make it efficient for this type of workload.

  • 人工智能/机器学习 (AI/ML)Artificial intelligence/machine learning (AI/ML). AI/ML 可以解决不同数据类型(例如视觉对象、语音和文本)的使用和处理。AI/ML deals with the consumption and processing of different data types like visuals, speech, and text. 这种高性能计算类型的工作负荷可以处理大量的、需要快速做出响应并在短时间内引入以进行高效数据分析的数据。This high-performance computing type of workload deals with large amounts of data that requires rapid response and efficient ingestion times for data analysis.

  • 数据转换Data transformation. 需要不断编辑、修改和转换数据的过程需要即时更新。Processes that require constant editing, modification, and conversion of data require instant updates. 为了准确表示数据,此数据的使用者必须立即看到这些更改。For accurate data representation, consumers of this data must see these changes reflected immediately.

标准性能Standard performance

标准性能支持不同的访问层以最经济高效的方式存储数据。Standard performance supports different access tiers to store data in the most cost-effective manner. 它已针对大型数据集的高容量和高吞吐量进行了优化。It's optimized for high capacity and high throughput on large data sets.

  • 备份和灾难恢复数据集Backup and disaster recovery datasets. 标准性能存储提供经济高效的层,使其成为短期和长期灾难恢复数据集、辅助备份与合规性数据存档的完美用例。Standard performance storage offers cost-efficient tiers, making it a perfect use case for both short-term and long-term disaster recovery datasets, secondary backups, and compliance data archiving.

  • 媒体内容Media content. 首次创建并存储图像和视频后,往往会频率地对其进行访问,但一段时间后,对此类内容的使用会越来越少。Images and videos often are accessed frequently when first created and stored, but this content type is used less often as it gets older. 标准性能存储提供适当的层来满足媒体内容需求。Standard performance storage offers suitable tiers for media content needs.

  • 批量数据处理Bulk data processing. 此类工作负荷适合使用标准存储,因为它们需要经济高效的高吞吐量存储,而不是一致的低延迟。These kinds of workloads are suitable for standard storage because they require cost-effective high-throughput storage instead of consistent low latency. 较大的原始数据集将会暂存以进行处理,最终会迁移到较冷的层。Large, raw datasets are staged for processing and eventually migrate to cooler tiers.

从标准层迁移到高级层Migrate from standard to premium

无法将现有的标准性能存储帐户转换为使用高级性能层的块 Blob 存储帐户。You can't convert an existing standard performance storage account to a block blob storage account with premium performance. 若要迁移到高级性能存储帐户,必须创建一个块 Blob 存储帐户,并将数据迁移到新帐户。To migrate to a premium performance storage account, you must create a BlockBlobStorage account, and migrate the data to the new account. 有关详细信息,请参阅创建块 Blob 存储帐户For more information, see Create a BlockBlobStorage account.

若要在存储帐户之间复制 Blob,可以使用最新版本的 AzCopy 命令行工具。To copy blobs between storage accounts, you can use the latest version of the AzCopy command-line tool. Azure 数据工厂等其他工具也可用于数据移动和转换。Other tools such as Azure Data Factory are also available for data movement and transformation.

Blob 生命周期管理Blob lifecycle management

Blob 存储生命周期管理提供了丰富的基于规则的策略:Blob storage lifecycle management offers a rich, rule-based policy:

  • Premium:在数据生命周期结束时使数据过期。Premium: Expire data at the end of its lifecycle.
  • 标准:将数据转换到最佳访问层,并在其生命周期结束时使数据过期。Standard: Transition data to the best access tier and expire data at the end of its lifecycle.

若要了解详细信息,请参阅管理 Azure Blob 存储生命周期To learn more, see Manage the Azure Blob storage lifecycle.

无法在热、冷和存档层之间移动存储在高级块 Blob 存储帐户中的数据。You can't move data that's stored in a premium block blob storage account between hot, cool, and archive tiers. 但是,可将块 Blob 存储帐户中的 Blob 复制到不同帐户中的热访问层。However, you can copy blobs from a block blob storage account to the hot access tier in a different account. 若要将数据复制到另一帐户,请使用从 URL 放置块 API 或 AzCopy v10To copy data to a different account, use the Put Block From URL API or AzCopy v10. “从 URL 放置块”API 以同步方式复制服务器上的数据。The Put Block From URL API synchronously copies data on the server. 仅在已将所有数据从原始服务器位置移到目标位置后,调用才会完成。The call completes only after all the data is moved from the original server location to the destination location.

后续步骤Next steps

评估 GPv2 和 Blob 存储帐户中的热、冷和存档层。Evaluate hot, cool, and archive in GPv2 and Blob storage accounts.