關於 Azure SQL 超大規模資料庫的常見問題FAQ about Azure SQL Hyperscale databases

本文針對考慮 Azure SQL Database 超大規模資料庫服務層級中的資料庫 (通常稱為超大規模資料庫資料庫) 的客戶, 提供常見問題的解答。This article provides answers to frequently asked questions for customers considering a database in the Azure SQL Database Hyperscale service tier, commonly called a Hyperscale database. 本文將說明「超大規模資料庫」支援的案例,以及在一般情況下與「SQL Database 超大規模資料庫」相容的跨功能服務。This article describes the scenarios that Hyperscale supports and the cross-feature services are compatible with SQL Database Hyperscale in general.

  • 此常見問題集適用於大致了解超大規模資料庫服務層級,而想就其特定的問題與顧慮尋求解答的讀者。This FAQ is intended for readers who have a brief understanding of the Hyperscale service tier and are looking to have their specific questions and concerns answered.
  • 此常見問題集並非 SQL Database 超大規模資料庫的使用指南,也不會解答關於操作方式的問題。This FAQ isn’t meant to be a guidebook or answer questions on how to use a SQL Database Hyperscale database. 如有此需求,建議您參閱 Azure SQL Database 超大規模資料庫文件。For that, we recommend you refer to the Azure SQL Database Hyperscale documentation.

一般問題General questions

什麼是超大規模資料庫What is a Hyperscale database

超大規模資料庫是超大規模資料庫服務層級之中採用超大規模資料庫相應放大儲存體技術的 Azure SQL 資料庫。A Hyperscale database is an Azure SQL database in the Hyperscale service tier that is backed by the Hyperscale scale-out storage technology. 超大規模資料庫最多可支援 100 TB 的資料,並提供高輸送量和高效能,以及快速調整以因應工作負載需求的能力。A Hyperscale database supports up to 100 TB of data and provides high throughput and performance, as well as rapid scaling to adapt to the workload requirements. 調整對應用程式而言是透明的 (連線性、查詢處理等等),運作方式與任何其他 SQL 資料庫相同。Scaling is transparent to the application – connectivity, query processing, and so on, work like any other SQL database.

哪些資源類型和購買模型支援超大規模資料庫What resource types and purchasing models support Hyperscale

超大規模資料庫服務層級僅適用於 Azure SQL Database 中的單一資料庫,且採用的必須是以虛擬核心為基礎的購買模型。The Hyperscale service tier is only available for single databases using the vCore-based purchasing model in Azure SQL Database.

超大規模資料庫服務層級與一般用途和商務關鍵服務層級有何不同How does the Hyperscale service tier differ from the General Purpose and Business Critical service tiers

以 vCore 為基礎的服務層級主要是根據可用性、儲存體類型和 IOPs 來區分。The vCore-based service tiers are primarily differentiated based upon availability, storage type, and IOPs.

  • 一般用途服務層級針對 IO 延遲或容錯移轉時間並非優先考量的案例提供了一組平衡的計算和儲存體選項,適用於大多數的商務工作負載。The General Purpose service tier is appropriate for most business workloads, offering a balanced set of compute and storage options where IO latency or failover times are not the priority.
  • 超大規模資料庫服務層級對於非常大的資料庫工作負載可發揮最佳效益。The Hyperscale service tier is optimized for very large database workloads.
  • 商務關鍵服務層級適用於 IO 延遲是優先考量的商務工作負載。The Business Critical service tier is appropriate for business workloads where IO latency is a priority.
資源類型Resource type 一般目的General Purpose 超大規模Hyperscale 商務關鍵性Business Critical
適用對象Best for 全部All 大部分的商業工作負載。Most business workloads. 提供以預算為導向且平衡的計算與儲存體選項。Offers budget oriented balanced compute and storage options. 具有大型資料容量需求,且能夠自動調整儲存體並流暢地調整計算的資料應用程式。Data applications with large data capacity requirements and the ability to auto-scale storage and scale compute fluidly. 具有高交易率和最低延遲 IO 的 OLTP 應用程式。OLTP applications with high transaction rate and lowest latency IO. 使用數個分開的複本,針對失敗提供最高的復原能力。Offers highest resilience to failures using several, isolated replicas.
資源類型Resource type 單一資料庫/彈性集區/受控執行個體Single database / elastic pool / managed instance 單一資料庫Single database 單一資料庫/彈性集區/受控執行個體Single database / elastic pool / managed instance
計算大小Compute size 單一資料庫/彈性集區 *Single database / elastic pool * 1 到 80 個虛擬核心1 to 80 vCores 1 到 80 個虛擬核心*1 to 80 vCores* 1 到 80 個虛擬核心1 to 80 vCores
受管理的執行個體Managed instance 8 個、16 個、24 個、32 個、40 個、64 個、80 個虛擬核心8, 16, 24, 32, 40, 64, 80 vCores N/AN/A 8 個、16 個、24 個、32 個、40 個、64 個、80 個虛擬核心8, 16, 24, 32, 40, 64, 80 vCores
儲存體類型Storage type 全部All 進階遠端儲存體 (每個執行個體)Premium remote storage (per instance) 與本機 SSD 快取分離的儲存體 (每個執行個體)De-coupled storage with local SSD cache (per instance) 超快速本機 SSD 儲存體 (每個執行個體)Super-fast local SSD storage (per instance)
儲存體大小Storage size 單一資料庫/彈性集區Single database / elastic pool 5 GB – 4 TB5 GB – 4 TB 最多 100 TBUp to 100 TB 5 GB – 4 TB5 GB – 4 TB
受管理的執行個體Managed instance 32 GB – 8 TB32 GB – 8 TB N/AN/A 32 GB – 4 TB32 GB – 4 TB
IO 輸送量IO throughput 單一資料庫**Single database** 每個虛擬核心 500 IOPS,且 IOPS 上限為 7000500 IOPS per vCore with 7000 maximum IOPS 超大規模資料庫是多層式架構, 在多個層級進行快取。Hyperscale is a multi-tiered architecture with caching at multiple levels. 有效的 IOPs 將視工作負載而定。Effective IOPs will depend on the workload. 5000 IOPS,IOPS 上限為 200,0005000 IOPS with 200,000 maximum IOPS
受管理的執行個體Managed instance 視檔案大小而定Depends on size of file N/AN/A 受控執行個體:視檔案大小而定Managed Instance: Depends on size of file
AvailabilityAvailability 全部All 1 個複本、無讀取規模、無本機快取1 replica, no read-scale, no local cache 多個複本、最多 15 個讀取規模、部分本機快取Multiple replicas, up to 15 read-scale, partial local cache 3 個複本、1 個讀取規模、區域備援 HA、完整本機快取3 replicas, 1 read-scale, zone-redundant HA, full local cache
備份Backups 全部All RA-GRS、7-35 天 (預設為 7 天)RA-GRS, 7-35 days (7 days by default) GRS、7天、固定時間點恢復 (PITR)RA-GRS, 7 days, constant time point-in-time recovery (PITR) RA-GRS、7-35 天 (預設為 7 天)RA-GRS, 7-35 days (7 days by default)

* 超大規模資料庫服務層級不支援彈性集區* Elastic pools not supported in the Hyperscale service tier

誰應使用超大規模資料庫服務層級Who should use the Hyperscale service tier

超大規模資料庫服務層級主要適用於具有大型內部部署 SQL Server 資料庫,且想要藉由移至雲端來現代化其應用程式的客戶,或是已使用 SQL Server 資料庫而想要大幅擴充資料庫成長潛能的客戶。The Hyperscale service tier is primarily intended for customers who have large on-premises SQL Server databases and want to modernize their applications by moving to the cloud or for customers who are already using Azure SQL Database and want to significantly expand the potential for database growth. 超大規模資料庫也適用於想要兼顧高效能和高延展性的客戶。Hyperscale is also intended for customers who seek both high performance and high scalability. 使用超大規模資料庫,您將可:With Hyperscale, you get:

  • 支援最多 100 TB 的資料庫大小Support for up to 100 TB of database size
  • 快速備份資料庫而無須考慮資料庫大小 (備份以檔案快照集為基礎)Fast database backups regardless of database size (backups are based on file snapshots)
  • 快速還原資料庫而無須考慮資料庫大小 (還原會從檔案快照集執行)Fast database restores regardless of database size (restores are from file snapshots)
  • 提高記錄輸送量進而加快交易認可速度,而無須考慮資料庫大小Higher log throughput results in fast transaction commit times regardless of database size
  • 將讀取規模相應放大至一或多個唯讀節點,以卸除讀取工作負載,並作為熱待命。Read scale out to one or more read-only nodes to offload your read workload, and for hot-standbys.
  • 快速相應增加計算 (依常數時間) 以充分因應大量工作負載中,然後再依常數時間相應減少。Rapid scale up of compute, in constant time, to be more powerful to accommodate the heavy workload and then scale down, in constant time. 其運作方式類似於 P6 到 P11 之間的相應增加和相應減少,但速度會快得多,因為這並非資料作業的大小。This is similar to scaling up and down between a P6 to a P11, for example, but much faster as this is not a size of data operation.

目前有哪些區域支援超大規模資料庫What regions currently support Hyperscale

Azure SQL Database 的超大規模資料庫層目前可在Azure SQL Database 超大規模資料庫總覽底下列出的區域中取得。The Azure SQL Database Hyperscale tier is currently available in the regions listed under Azure SQL Database Hyperscale Overview.

是否可為每個邏輯伺服器建立多個超大規模資料庫Can I create multiple Hyperscale databases per logical server

是的。Yes. 如需每個邏輯伺服器的超大規模資料庫數目限制的詳細資訊,請參閱適用於邏輯伺服器上之單一和集區資料庫的 SQL Database 資源限制For more information and limits on the number of Hyperscale databases per logical server, see SQL Database resource limits for single and pooled databases on a logical server.

超大規模資料庫資料庫的效能特性為何What are the performance characteristics of a Hyperscale database

SQL Database 超大規模資料庫架構不僅支援大型資料庫,同時也可提供高效能和高輸送量。The SQL Database Hyperscale architecture provides high performance and throughput while supporting large database sizes.

超大規模資料庫有何延展性What is the scalability of a Hyperscale database

「SQL Database 超大規模資料庫」可根據您的工作負載需求提供快速延展性。SQL Database Hyperscale provides rapid scalability based on your workload demand.

  • 相應增加/減少Scaling Up/Down

    使用超大規模資料庫時,您可以在 CPU、記憶體等資源方面相應增加主要計算大小,然後再依常數時間相應減少。With Hyperscale, you can scale up the primary compute size in terms of resources like CPU, memory and then scale down, in constant time. 由於儲存體是共用的,因此相應增加和相應減少並非資料作業的大小。Because the storage is shared, scaling up and scaling down is not a size of data operation.

  • 相應縮小/相應放大Scaling In/Out

    使用超大規模資料庫時,您也將能夠佈建一或多個額外的計算節點,用來支應您的讀取要求。With Hyperscale, you also get the ability to provision one or more additional compute nodes that you can use to serve your read requests. 這表示,您可以將這些額外的計算節點作為唯讀節點,用以卸除主要計算節點的讀取工作負載。This means that you can use these additional compute nodes as read-only nodes to offload your read workload from the primary compute. 除了唯讀以外, 這些節點也會在容錯移轉主要時作為熱待命。In addition to read-only, these nodes also serve as hot-standby’s in the event of a failover from the primary.

    佈建每個這些額外計算節點的作業也可在常數時間內完成,而且這是線上作業。Provisioning of each of these additional compute nodes can be done in constant time and is an online operation. 您可以將連接字串的 ApplicationIntent 引數設為 readonly,藉以連線至這些額外的唯讀計算節點。You can connect to these additional read-only compute nodes by setting the ApplicationIntent argument on your connection string to readonly. 任何標示為 readonly 的連線都會自動路由至其中一個額外的唯讀計算節點。Any connections marked with readonly are automatically routed to one of the additional read-only compute nodes.

深入問題Deep dive questions

我可以在單一邏輯伺服器中混用超大規模資料庫與單一資料庫Can I mix Hyperscale and single databases in a single logical server

是,您可以這麼做。Yes, you can.

使用「超大規模資料庫」時是否需要變更應用程式設計模型Does Hyperscale require my application programming model to change

否,您的應用程式設計模型應保持原狀。No, your application programming model stays as is. 您應如常使用連接字串,並使用其他一般模式與 Azure SQL 資料庫進行互動。You use your connection string as usual and the other regular modes to interact with your Azure SQL database.

SQL Database 超大規模資料庫上會有哪些預設的交易隔離等級What transaction isolation levels are going to be default on SQL Database Hyperscale database

主要節點上的交易隔離等級是 RCSI (讀取認可快照集隔離)。On the primary node, the transaction isolation level is RCSI (Read Committed Snapshot Isolation). 在讀取規模次要節點上,隔離等級是快照集。On the read scale secondary nodes, the isolation level is Snapshot.

內部部署或 IaaS SQL Server 授權是否可用在 SQL Database 超大規模資料庫上Can I bring my on-premises or IaaS SQL Server license to SQL Database Hyperscale

是,Azure Hybrid Benefit 適用於「超大規模資料庫」。Yes, Azure Hybrid Benefit is available for Hyperscale. 每個 SQL Server Standard 核心可對應至 1 個超大規模資料庫虛擬核心。Every SQL Server Standard core can map to 1 Hyperscale vCores. 每個 SQL Server Enterprise 核心可對應至 4 個超大規模資料庫虛擬核心。Every SQL Server Enterprise core can map to 4 Hyperscale vCores. 次要複本不需要 SQL 授權。You don’t need a SQL license for secondary replicas. Azure Hybrid Benefit 價格會自動套用至讀取規模 (次要) 複本。The Azure Hybrid Benefit price will be automatically applied to read-scale (secondary) replicas.

SQL Database 超大規模資料庫是針對何種工作負載而設計的What kind of workloads is SQL Database Hyperscale designed for

「SQL Database 超大規模資料庫」支援所有 SQL Server 工作負載,但主要針對 OLTP 進行最佳化。SQL Database Hyperscale supports all SQL Server workloads, but it is primarily optimized for OLTP. 您也可以帶入混合式 (HTAP) 和分析 (資料超市) 工作負載。You can bring Hybrid (HTAP) and Analytical (data mart) workloads as well.

如何在 Azure SQL 資料倉儲與 SQL Database 超大規模資料庫之間做選擇How can I choose between Azure SQL Data Warehouse and SQL Database Hyperscale

如果您目前使用 SQL Server 做為資料倉儲來執行互動式分析查詢, SQL Database 超大規模資料庫是很好的選擇, 因為您可以更低的成本裝載相對較小的資料倉儲 (例如, 10s TB 的幾 TB), 而且您可以將資料移轉至將工作負載 arehouse 至 SQL Database 超大規模資料庫, 而不需要 T-sql 程式碼變更。If you are currently running interactive analytics queries using SQL Server as a data warehouse, SQL Database Hyperscale is a great option because you can host relatively small data warehouses (such as a few TB up to 10s of TB) at a lower cost and you can migrate your data warehouse workload to SQL Database Hyperscale without T-SQL code changes.

如果您使用平行處理資料倉儲 (PDW)、Teradata 或其他巨量平行處理器 (MPP) 資料倉儲,以複雜的查詢大規模執行資料分析,則 SQL 資料倉儲將是最佳選擇。If you are running data analytics on a large scale with complex queries and using Parallel Data Warehouse (PDW), Teradata or other Massively Parallel Processor (MPP)) data warehouses, SQL Data Warehouse may be the best choice.

SQL Database 超大規模資料庫的計算問題SQL Database Hyperscale compute questions

我是否可隨時暫停計算Can I pause my compute at any time

不過, 此時您可以相應縮小計算和數目的複本, 以降低非尖峰時間的成本。Not at this time, however you can scale your compute and number of replicas down to reduce cost during non-peak times.

我是否可為記憶體密集工作負載佈建額外增加 RAM 的計算Can I provision a compute with extra RAM for my memory-intensive workload

資料分割No. 若要有更多的 RAM,您需要升級至更高的計算大小。To get more RAM, you need to upgrade to a higher compute size. 如需詳細資訊,請參閱超大規模資料庫儲存體和計算大小For more information, see Hyperscale storage and compute sizes.

我是否可以佈建不同大小的多個計算節點Can I provision multiple compute nodes of different sizes

資料分割No.

可支援多少個讀取規模複本How many read-scale replicas are supported

預設會使用一個讀取規模複本 (總共兩個複本) 來建立超大規模資料庫資料庫。The Hyperscale databases are created with one read-scale replica (two replicas in total) by default. 您可以使用Azure 入口網站t-sqlPowershellCLI, 來調整0和4之間的唯讀複本數目。You can scale the number of read-only replicas between 0 and 4 using the Azure portal, T-SQL, Powershell or CLI.

是否需要佈建額外的計算節點才能取得高可用性For high availability, do I need to provision additional compute nodes

在超大規模資料庫資料庫中, 會在儲存體層級提供復原功能。In Hyperscale databases, the resiliency is provided at the storage level. 您只需要一個複本來提供復原功能。You only need one replica to provide resiliency. 計算複本失效時,將會自動建立新複本,且不會遺失資料。When the compute replica is down, a new replica is created automatically with no data loss.

但如果只有一個複本,則在容錯移轉之後可能需要一些時間才能在新複本中建置本機快取。However, if there’s only one replica, it may take some time to build the local cache in the new replica after failover. 在快取重建階段中,資料庫會直接從頁面伺服器提取資料,而導致 IOPS 和查詢效能降低。During the cache rebuild phase, the database fetches data directly from the page servers, resulting in degraded IOPS and query performance.

針對需要高可用性的關鍵任務應用程式,您應佈建至少 2 個計算節點,包括主要計算節點在內 (預設值)。For mission-critical apps that require high availability, you should provision at least 2 compute nodes including the primary compute node (default). 如此,在執行容錯移轉時即有熱待命節點可供使用。That way there is a hot-standby available in the case of a failover.

資料大小和儲存體問題Data size and storage questions

「SQL Database 超大規模資料庫」最多可支援多大的資料庫What is the max db size supported with SQL Database Hyperscale

100 TB100 TB

「超大規模資料庫」的交易記錄大小為何What is the size of the transaction log with Hyperscale

「超大規模資料庫」的交易記錄可說是無限制的。The transaction log with Hyperscale is practically infinite. 您無須擔心記錄輸送量偏高的系統上是否會有記錄空間不足的問題。You do not need to worry about running out of log space on a system that has a high log throughput. 不過,如果工作負載持續湧現,記錄產生速率可能會受到節流控制。However, the log generation rate might be throttled for continuous aggressive workloads. 尖峰持續記錄產生速率大約是 100 MB/秒。The peak sustained log generation rate is approximately 100 MB/sec.

暫存資料庫是否會隨著資料庫成長而調整Does my temp db scale as my database grows

您的 tempdb 資料庫位於本機 SSD 儲存體,會根據您佈建的計算大小設定。Your tempdb database is located on local SSD storage and is configured based on the compute size that you provision. 您的 tempdb 已進行最佳化和配置,可提供最大效能優勢。Your tempdb is optimized and laid out to provide maximum performance benefits. tempdb 大小是不可設定的,儲存體子系統會替您加以管理。The tempdb size is not configurable and is managed for you by storage sub-system.

資料庫大小會自動成長,還是必須由我管理資料檔案的大小Does my database size automatically grow, or do I have to manage the size of the data files

您的資料庫大小會在您插入/擷取更多資料時自動成長。Your database size automatically grows as you insert/ingest more data.

SQL Database 超大規模資料庫支援或一開始的最小資料庫大小為何What is the smallest database size that SQL Database Hyperscale supports or starts with

10 GB10 GB

資料庫大小成長的遞增量為何In what increments does my database size grow

1 GB1 GB

「SQL Database 超大規模資料庫」中的儲存體是本機還是遠端的Is the storage in SQL Database Hyperscale local or remote

在「超大規模資料庫」中,資料檔案會儲存在 Azure 標準儲存體中。In Hyperscale, data files are stored in Azure standard storage. 資料會大量快取到計算節點近端電腦的本機 SSD 儲存體上。Data is heavily cached on local SSD storage, on machines close to the compute nodes. 此外,計算節點會有本機 SSD 和記憶體內部 (緩衝集區等等) 的快取,以降低從遠端節點提取資料的頻率。In addition, compute nodes have a cache on local SSD and in-memory (Buffer Pool, and so on), to reduce the frequency of fetching data from remote nodes.

我是否可管理或定義「超大規模資料庫」的檔案或檔案群組Can I manage or define files or filegroups with Hyperscale

No

我是否可為資料庫佈建資料成長的強制上限Can I provision a hard cap on the data growth for my database

No

「SQL Database 超大規模資料庫」會如何配置資料檔案How are data files laid out with SQL Database Hyperscale

資料檔案由頁面伺服器所控制。The data files are controlled by page servers. 當資料大小成長時,即會新增資料檔案和相關聯的頁面伺服器節點。As the data size grows, data files and associated page server nodes are added.

是否支援資料庫縮減Is database shrink supported

No

是否支援資料庫壓縮Is database compression supported

Yes

如果我有很大的資料表,資料表的資料是否會分散到多個資料檔案If I have a huge table, does my table data get spread out across multiple data files

是的。Yes. 與指定的資料表相關聯的資料頁面可能會移至多個資料檔案中,但全都屬於相同的檔案群組。The data pages associated with a given table can end up in multiple data files, which are all part of the same filegroup. SQL Server 會使用按比例填滿策略將資料分散到多個資料檔案。SQL Server uses a proportional fill strategy to distribute data over data files.

資料移轉問題Data migration questions

是否可將現有的 Azure SQL 資料庫移至超大規模資料庫服務層級Can I move my existing Azure SQL databases to the Hyperscale service tier

是的。Yes. 您可以將現有的 Azure SQL 資料庫移至「超大規模資料庫」。You can move your existing Azure SQL databases to Hyperscale. 這是單向的遷移。This is a one-way migration. 您無法將資料庫從超大規模資料庫移到另一個服務層級中。You can’t move databases from Hyperscale to another service tier. 建議您建立一份生產資料庫副本,並遷移至超大規模資料庫移轉以進行概念證明 (POC)。We recommend you make a copy of your production databases and migrate to Hyperscale for proof of concepts (POCs).

是否可將超大規模資料庫移至其他版本Can I move my Hyperscale databases to other editions

資料分割No. 目前, 您無法將超大規模資料庫資料庫移至另一個服務層級。At this time, you can’t move a Hyperscale database to another service tier.

移轉至超大規模資料庫服務層級之後是否會失去任何運作性或功能Do I lose any functionality or capabilities after migration to the Hyperscale service tier

是的。Yes. 超大規模資料庫尚不支援某些 Azure SQL Database 功能, 包括但不限於長期保留備份。Some of Azure SQL Database features are not supported in Hyperscale yet, including but not limited long term retention backup. 在您將資料庫移轉至「超大規模資料庫」後,這些功能將會停止運作。After you migrate your databases to Hyperscale, those features stop working. 我們希望這些限制是暫時性的。We expect these limitations to be temporary.

我是否可將內部部署 SQL Server 資料庫或 SQL Server 虛擬機器資料庫移至「超大規模資料庫」Can I move my on-premises SQL Server database or my SQL Server virtual machine database to Hyperscale

是的。Yes. 您可以使用所有現有的移轉技術來移轉至「超大規模資料庫」,包括 BACPAC、異動複寫、邏輯資料載入等。You can use all existing migration technologies to migrate to Hyperscale, including BACPAC, transactional replication, logical data loading. 另請參閱 Azure 資料庫移轉服務See also the Azure Database Migration Service.

從內部部署或虛擬機器環境移轉至「超大規模資料庫」期間的停機時間有多長?如何盡量縮短What is my downtime during migration from an on-premises or virtual machine environment to Hyperscale and how can I minimize it

此停機時間與您將資料庫移轉至 Azure SQL Database 中的單一資料庫時產生的停機時間相同。Downtime is the same as the downtime when you migrate your databases to a single database in Azure SQL Database. 移轉大小未超過 10 TB 的資料庫時,您可以使用異動複寫盡可能縮短停機時間。You can use transactional replication to minimize downtime migration for databases up to few TB in size. 針對非常大型的資料庫 (10+ TB),則應考慮使用 ADF、Spark 或其他資料移動技術來移轉資料。For very large database (10+ TB), you can consider to migrate data using ADF, Spark, or other data movement technologies.

將某個數量的資料移至「SQL Database 超大規模資料庫」需要多少時間How much time would it take to bring in X amount of data to SQL Database Hyperscale

超大規模資料庫能夠耗用 100 MB/秒的新/已變更資料。Hyperscale is capable of consuming 100 MB/sec of new/changed data.

我是否可從 Blob 儲存體讀取資料並執行快速載入 (例如 Polybase 和 SQL 資料倉儲)Can I read data from blob storage and do fast load (like Polybase and SQL Data Warehouse)

您可以從 Azure 儲存體讀取資料,並將資料載入超大規模資料庫中 (就像使用一般的單一資料庫一樣)。You can read data from Azure Storage and load data load into a Hyperscale database (just like you can do with a regular single database). Azure SQL Database 目前不支援 PolyBase。Polybase is currently not supported on Azure SQL Database. 您可以使用 Azure Data Factory,或透過 SQL 的 Spark 連接器Azure Databricks 中執行 Spark 作業,藉以執行 Polybase。You can do Polybase using Azure Data Factory or running a Spark job in Azure Databricks with the Spark connector for SQL. SQL 的 Spark 連接器支援大量插入。The Spark connector to SQL supports bulk insert.

「超大規模資料庫」不支援簡單復原或大量記錄模式。Simple recovery or bulk logging model is not supported in Hyperscale. 若要提供高可用性,必須使用完整復原模式。Full recovery model is required to provide high availability. 不過,「超大規模資料庫」憑藉新的記錄架構,而可提供優於單一 Azure SQL 資料庫的資料擷取速率。However, Hyperscale provides a better data ingest rate compared to a single Azure SQL database because of the new log architecture.

「SQL Database 超大規模資料庫」是否允許佈建多個節點以擷取大量資料Does SQL Database Hyperscale allow provisioning multiple nodes for ingesting large amounts of data

資料分割No. 「SQL Database 超大規模資料庫」屬於 SMP 架構,而不是非對稱多工處理或多重主機架構。SQL Database Hyperscale is a SMP architecture and is not an asymmetric multiprocessing or a multi-master architecture. 您只能建立多個複本以相應放大唯讀工作負載。You can only create multiple replicas to scale out read-only workloads.

可移轉至「SQL Database 超大規模資料庫」的 SQL Server 最舊版本為何What is the oldest SQL Server version will SQL Database Hyperscale support migration from

SQL Server 2005。SQL Server 2005. 如需詳細資訊,請參閱移轉至單一資料庫或集區資料庫For more information, see Migrate to a single database or a pooled database. 若要了解相容性問題,請參閱解決資料庫移轉相容性問題For compatibility issues, see Resolving database migration compatibility issues.

「SQL Database 超大規模資料庫」是否支援從其他資料來源進行移轉,例如 Aurora、MySQL、Oracle、DB2 和其他資料庫平台Does SQL Database Hyperscale support migration from other data sources such as Aurora, MySQL, Oracle, DB2, and other database platforms

是的。Yes. 如果資料來源不是 SQL Server,則需進行邏輯移轉。Coming from different data sources other than SQL Server requires logical migration. 您可以使用 Azure 資料庫移轉服務進行邏輯移轉。You can use the Azure Database Migration Service for a logical migration.

商務持續性和災害復原問題Business continuity and disaster recovery questions

超大規模資料庫隨附的 SLA 為何What SLA’s are provided for a Hyperscale database

預設的主要複本加1可讀取的次要複本, SLA 為 99.95% 的可用性。With the default primary plus 1 readable secondary, the SLA is 99.95% availability. 有更多複本, SLA 的最高可達 99.99%。With more replicas, the SLA goes up to 99.99%.

Azure SQL Database 服務是否會為我管理資料庫備份Are the database backups managed for me by the Azure SQL Database service

Yes

資料庫備份多久執行一次How often are the database backups taken

SQL Database 超大規模資料庫並沒有傳統的完整、差異和記錄備份。There are no traditional full, differential, and log backups for SQL Database Hyperscale databases. 它會有資料檔案的一般快照集,而產生的記錄只會依原狀在設定的保留期限內保存,或供您使用。Instead, there are regular snapshots of the data files and log that is generated is simply retained as is for the retention period configured or available to you.

「SQL Database 超大規模資料庫」是否支援時間點還原Does SQL Database Hyperscale support Point in Time Restore

Yes

「SQL Database 超大規模資料庫」中的備份/還原具有怎樣的復原點目標 (RPO)/復原時間目標 (RTO)What is the Recovery Point Objective (RPO)/Recovery Time Objective (RTO) with backup/restore in SQL Database Hyperscale

RPO 為 0 分鐘。RTO 目標小於 10 分鐘,無論資料庫大小為何。The RPO is 0 min. The RTO goal is less than 10 minutes, regardless of database size.

大型資料庫的備份是否會影響主要資料庫的計算效能Do backups of large databases affect compute performance on my primary

資料分割No. 備份由儲存子系統管理,並使用檔案快照集。Backups are managed by the storage subsystem, and leverage file snapshots. 這並不會影響到主要資料庫上的使用者工作負載。They do not impact the user workload on the primary.

是否可使用 SQL Database 超大規模資料庫執行異地還原Can I perform geo-restore with a SQL Database Hyperscale database

是的。Yes. 完全支援異地還原。Geo-restore is fully supported.

是否可使用 SQL Database 超大規模資料庫設定異地複寫Can I setup Geo-Replication with SQL Database Hyperscale database

目前沒有。Not at this time.

我的第二個計算節點是否可使用「SQL Database 超大規模資料庫」進行異地複寫Do my secondary compute nodes get geo-replicated with SQL Database Hyperscale

目前沒有。Not at this time.

是否可建立 SQL Database 超大規模資料庫備份,而後將其還原至內部部署伺服器或 VM 中 SQL ServerCan I take a SQL Database Hyperscale database backup and restore it to my on-premises server or SQL Server in VM

資料分割No. 超大規模資料庫的儲存格式會與傳統 SQL Server 不同,且您無法控制備份或加以存取。The storage format for Hyperscale databases is different from traditional SQL Server, and you don’t control backups or have access to them. 若要從 SQL Database 超大規模資料庫中取出您的資料,請使用匯出服務,或搭配使用指令碼和 BCP。To take your data out of a SQL Database Hyperscale database, either use the export service or use scripting plus BCP.

跨功能問題Cross Feature questions

移轉至超大規模資料庫服務層級之後是否會失去任何運作性或功能Do I lose any functionality or capabilities after migration to the Hyperscale service tier

是的。Yes. 某些 Azure SQL Database 功能在超大規模資料庫中不受支援, 包括但不限於長期保留備份。Some of Azure SQL Database features are not supported in Hyperscale, including but not limited long term retention backup. 在您將資料庫移轉至「超大規模資料庫」後,這些功能將會停止運作。After you migrate your databases to Hyperscale, those features stop working.

Polybase 適用於「SQL Database 超大規模資料庫」Will Polybase work with SQL Database Hyperscale

資料分割No. Azure SQL Database 不支援 PolyBase。Polybase isn’t supported on Azure SQL Database.

計算是否支援 R 和 PythonDoes the compute have support for R and python

資料分割No. Azure SQL Database 不支援 R 和 Python。R and Python are not supported in Azure SQL Database.

計算節點是否會容器化Are the compute nodes containerized

資料分割No. 您的資料庫位於計算 VM 上,而不是容器中。Your database resides on a compute VM and not a container.

效能問題Performance questions

在最大的 SQL Database 超大規模資料庫計算上最多可推送多少輸送量How much throughput can I push on the largest SQL Database Hyperscale compute

我們看到變更資料的 100 MB/秒一致 (交易記錄資料產生)We have seen a consistent 100 MB/sec of change data (transaction log data generation)

在最大的 SQL Database 超大規模資料庫計算上最多可達到多少 IOPSHow many IOPS do I get on the largest SQL Database Hyperscale compute

IOPS 和 IO 延遲會根據工作負載模式而有所不同。IOPS and IO latency will vary depending on the workload patterns. 如果需要存取的資料在計算的快取中是本機的, 它將會是與本機 SSD 相同的 IO 模式。If the data needing to be accessed is local to the compute's cache, it will be the same IO patterns as local SSD.

備份是否會影響到輸送量Does my throughput get affected by backups

資料分割No. 計算會分離於儲存層以外,以避免影響到計算。Compute is decoupled from the storage layer to avoid impact on compute.

在佈建額外的計算節點時是否會影響到輸送量Does my throughput get affected as I provision additional compute nodes

由於儲存體是共用的, 而且主要和次要計算節點之間不會進行直接的實體複寫, 因此在技術上, 主要節點上的輸送量不會受到新增讀取規模節點的影響。Because the storage is shared and there is no direct physical replication happening between primary and secondary compute nodes, technically, the throughput on primary node will not be affected by adding read-scale nodes. 不過,我們可以對持續湧現的工作負載進行節流控制,讓次要節點能夠套用記錄,並且讓頁面伺服器趕上進度,並避免次要節點上的讀取效能不佳。However, we may throttle continuous aggressive workload to allow log apply on secondary nodes and page servers to catch up, and avoid bad read performance on secondary nodes.

延展性問題Scalability questions

相應增加和相應減少計算節點需要多少時間How long would it take to scale up and down a compute node

無論資料大小為何, 相應增加或減少計算都應該花費5-10 分鐘的時間。Scaling compute up or down should take 5-10 minutes regardless of data size.

我的資料庫在相應增加/減少作業進行時是否會離線Is my database offline while the scaling up/down operation is in progress

資料分割No. 相應增加和相應減少會在線上執行。The scaling up and down will be online.

進行調整作業時是否可能發生連線中斷Should I expect connection drop when the scaling operations are in progress

如果具有目標大小的計算節點發生容錯移轉,相應增加或相應減少就會導致現有連線中斷。Scaling up or down results in existing connections being dropped when failover happens to the compute node with the target size. 新增讀取複本並不會導致連線中斷。Adding read replicas does not result in connection drops.

計算節點的相應增加和相應減少是自動作業還是使用者觸發的作業Is the scaling up and down of compute nodes automatic or end-user triggered operation

使用者。End-user. 並非自動的。Not automatic.

我的 tempb 是否會隨著計算的相應增加而成長Does my tempb also grow as the compute is scaled up

是的。Yes. 暫存資料庫將隨著計算的成長而自動相應增加。Temp db will scale up automatically as the compute grows.

我可以布建多個主要計算節點, 例如多個主要計算標頭可以驅動較高層級並行的多宿主系統Can I provision multiple primary compute nodes such as a multi-master system where multiple primary compute heads can drive a higher level of concurrency

資料分割No. 只有主要計算節點會接受讀取/寫入要求。Only the primary compute node accepts read/write requests. 次要計算節點僅接受唯讀要求。Secondary compute nodes only accept read-only requests.

讀取規模問題Read scale questions

我可以佈建多少個次要計算節點How many secondary compute nodes can I provision

我們預設會為超大規模資料庫資料庫建立2個複本。We create 2 replicas for Hyperscale databases by default. 如果您想要調整複本數目, 可以使用Azure 入口網站If you want to adjust the number of replicas, you can do so using Azure portal.

如何連線至這些次要計算節點How do I connect to these secondary compute nodes

您可以將連接字串的 ApplicationIntent 引數設為 readonly,藉以連線至這些額外的唯讀計算節點。You can connect to these additional read-only compute nodes by setting the ApplicationIntent argument on your connection string to readonly. 任何標示為 readonly 的連線都會自動路由至其中一個額外的唯讀計算節點。Any connections marked with readonly are automatically routed to one of the additional read-only compute nodes.

是否可以建立讀取規模複本的專用端點Can I create a dedicated endpoint for the read-scale replica

資料分割No. 您只能藉由指定ApplicationIntent=ReadOnly來連接到讀取級別複本。You can only connect to read-scale replica by specifying ApplicationIntent=ReadOnly.

系統是否會對讀取工作負載進行智慧型負載平衡Does the system do intelligent load balancing of the read workload

資料分割No. 唯讀工作負載會重新導向至隨機的讀取規模複本。The read-only workload is redirected to a random read-scale replica.

是否可在主要計算節點外獨立相應增加/減少次要計算節點Can I scale up/down the secondary compute nodes independently of the primary compute

資料分割No. 次要計算節點也會用於 HA, 因此在容錯移轉的情況下, 它們必須是與主要複本相同的設定。The secondary compute nodes are also used for HA, so they need to be the same configuration as the primary, in the case of a failover.

主要計算節點和其他次要計算節點是否可以有不同的暫存資料庫大小Do I get different temp db sizing for my primary compute and my additional secondary compute nodes

資料分割No. tempdb的會根據計算大小布建而設定, 次要計算節點的大小與主要計算相同。Your tempdb is configured based on the compute size provisioning, your secondary compute nodes are the same size as the primary compute.

是否可以在次要計算節點上新增索引和檢視Can I add indexes and views on my secondary compute nodes

資料分割No. 超大規模資料庫會共用儲存體,這表示所有計算節點會看到相同的資料表、索引和檢視。Hyperscale databases have shared storage, meaning that all compute nodes see the same tables, indexes and views. 如果您希望次要節點上有適用於讀取的額外索引 – 您必須在主要節點上加以新增。If you want additional indexes optimized for reads on secondary – you must add them on the primary first.

主要和次要計算節點之間會有多少延遲How much delay is there going to be between the primary and secondary compute node

從在主要節點上認可交易時起算,兩者幾乎沒有延遲,或只有數毫秒的延遲,視記錄產生速率而定。From the time a transaction is committed on the primary, depending on the log generation rate, it can either be instantaneous or in low milliseconds.

後續步驟Next steps

如需超大規模資料庫服務層級的詳細資訊, 請參閱超大規模資料庫服務層級For more information about the Hyperscale service tier, see Hyperscale service tier.