您现在访问的是微软AZURE全球版技术文档网站,若需要访问由世纪互联运营的MICROSOFT AZURE中国区技术文档网站,请访问 https://docs.azure.cn.

使用 Azure 顾问降低服务成本Reduce service costs by using Azure Advisor

通过识别闲置和未充分利用的资源,Azure 顾问可帮助优化和降低 Azure 总支出。Azure Advisor helps you optimize and reduce your overall Azure spend by identifying idle and underutilized resources. 可在顾问仪表板的“成本”选项卡获取成本建议。 You can get cost recommendations from the Cost tab on the Advisor dashboard.

通过调整或关闭未充分利用的实例来优化虚拟机花费Optimize virtual machine spend by resizing or shutting down underutilized instances

虽然某些应用程序方案有意使虚拟机利用率较低,但通过管理虚拟机大小和数量通常可降低成本。Although certain application scenarios can result in low utilization by design, you can often save money by managing the size and number of your virtual machines.

建议的操作是关机或重设大小,具体取决于要评估的资源。The recommended actions are shut down or resize, specific to the resource being evaluated.

当所有这些语句都为 true 时,顾问中的高级评估模型考虑关闭虚拟机:The advanced evaluation model in Advisor considers shutting down virtual machines when all of these statements are true:

  • CPU 利用率最大值的第 95 个百分位数小于 3%。P95th of the maximum of maximum value of CPU utilization is less than 3%.
  • 7 天内的网络利用率低于 2%。Network utilization is less than 2% over a seven-day period.
  • 内存压力低于阈值Memory pressure is lower than the threshold values

如果可以在较小 SKU(同一 SKU 系列内)或较少数量的实例中容纳当前负载,则顾问会考虑重设虚拟机大小,以使:Advisor considers resizing virtual machines when it's possible to fit the current load in a smaller SKU (within the same SKU family) or a smaller number of instances such that:

  • 不面向用户的工作负荷的当前负载利用率不超过 80%。The current load doesn’t go above 80% utilization for workloads that aren't user facing.
  • 面向用户的工作负荷的负载利用率不超过 40%。The load doesn't go above 40% for user-facing workloads.

在这里,顾问通过分析工作负荷的 CPU 利用率特征来确定工作负荷的类型。Here, Advisor determines the type of workload by analyzing the CPU utilization characteristics of the workload.

顾问会显示与建议的操作(重设大小或关机)相对应的成本节省估算值。Advisor shows the estimated cost savings for either recommended action: resize or shut down. 对于重设大小,顾问提供当前 SKU 信息和目标 SKU 信息。For resize, Advisor provides current and target SKU information.

若要加强对低利用率虚拟机的识别,可在每个订阅的基础上调整 CPU 利用率规则。If you want to be more aggressive about identifying underutilized virtual machines, you can adjust the CPU utilization rule on a per-subscription basis.

通过适当调整大小优化 MariaDB、MySQL 和 PostgreSQL 服务器的支出Optimize spend for MariaDB, MySQL, and PostgreSQL servers by right-sizing

顾问会分析使用情况并评估 MariaDB、MySQL 或 PostgreSQL 数据库服务器资源在过去 7 天内是否长时间未被充分利用。Advisor analyses your usage and evaluates whether your MariaDB, MySQL, or PostgreSQL database server resources have been underutilized for an extended time over the past seven days. 资源利用率低会导致不必要的支出,可在不显著影响性能的情况下解决这一问题。Low resource utilization results in unwanted expenditure that you can fix without significant performance impact. 若要降低成本并高效管理资源,我们建议将计算大小 (vCore) 减少一半。To reduce your costs and efficiently manage your resources, we recommend that you reduce the compute size (vCores) by half.

通过消除未设置的 ExpressRoute 线路来降低成本Reduce costs by eliminating unprovisioned ExpressRoute circuits

顾问会识别处于“未预配”提供程序状态超过一个月的 Azure ExpressRoute 线路。Advisor identifies Azure ExpressRoute circuits that have been in the provider status of Not provisioned for more than one month. 如果你不打算通过连接提供程序预配线路,则建议删除该线路。It recommends deleting the circuit if you aren't planning to provision the circuit with your connectivity provider.

通过删除或重新配置空闲虚拟网络网关来降低成本Reduce costs by deleting or reconfiguring idle virtual network gateways

顾问会识别已闲置超过 90 天的虚拟网关。Advisor identifies virtual network gateways that have been idle for more than 90 days. 因为这些网关按小时计费,所以如果不打算再使用它们,则应考虑重新配置或删除它们。Because these gateways are billed hourly, you should consider reconfiguring or deleting them if you don't intend to use them anymore.

购买虚拟机预留实例可节省即用即付成本Buy reserved virtual machine instances to save money over pay-as-you-go costs

顾问会查看过去 30 天的虚拟机使用情况,以确定是否可以通过购买 Azure 预留来节省成本。Advisor reviews your virtual machine usage over the past 30 days to determine if you could save money by purchasing an Azure reservation. 顾问会显示节省潜力最高的区域和大小,以及可通过购买预留实现的成本节省估算值。Advisor shows you the regions and sizes where the potential for savings is highest and the estimated savings from purchasing reservations. 通过 Azure 预留,可以预先购买虚拟机的基本成本。With Azure reservations, you can pre-purchase the base costs for your virtual machines. 折扣会自动应用于新的或现有的 VM,这些 VM 具有与预留相同的大小和区域。Discounts automatically apply to new or existing VMs that have the same size and region as your reservations. 深入了解 Azure 虚拟机预留实例。Learn more about Azure Reserved VM Instances.

顾问还会向你通知在接下来的 30 天内将过期的预留实例。Advisor also notifies you of your reserved instances that will expire in the next 30 days. 它会建议你购买新的预留实例来避免即用即付定价。It recommends that you purchase new reserved instances to avoid pay-as-you-go pricing.

购买多个资源类型的预留实例,以节省即用即付成本Buy reserved instances for several resource types to save over your pay-as-you-go costs

顾问会分析以下资源在过去 30 天内的使用模式,并建议可优化成本的预留容量购买量。Advisor analyzes usage patterns for the past 30 days for the following resources and recommends reserved capacity purchases that optimize costs.

Azure Cosmos DB 预留容量Azure Cosmos DB reserved capacity

顾问会分析 Azure Cosmos DB 在过去 30 天内的使用模式,并建议可优化成本的预留容量购买量。Advisor analyzes your Azure Cosmos DB usage patterns for the past 30 days and recommends reserved capacity purchases to optimize costs. 通过使用预留容量,可以预购买按小时计的 Azure Cosmos DB 使用量,从而节省即用即付成本。By using reserved capacity, you can pre-purchase Azure Cosmos DB hourly usage and save over your pay-as-you-go costs. 预留容量是一项计费权益,它会自动应用于新部署和现有部署。Reserved capacity is a billing benefit and automatically applies to new and existing deployments. 顾问基于 3 年期预留定价并根据过去 30 天内观察到的使用模式进行推测,从而计算各订阅的成本节省估算值。Advisor calculates savings estimates for individual subscriptions by using 3-year reservation pricing and by extrapolating the usage patterns observed over the past 30 days. 共享范围建议适用于预留容量购买,并可提高成本节省。Shared scope recommendations are available for reserved capacity purchases and can increase savings.

SQL 数据库和 SQL 托管实例保留容量SQL Database and SQL Managed Instance reserved capacity

顾问在过去30天内分析 SQL 数据库和 SQL 托管实例使用模式。Advisor analyzes SQL Database and SQL Managed Instance usage patterns over the past 30 days. 然后建议可优化成本的预留容量购买量。It then recommends reserved capacity purchases that optimize costs. 通过使用预留容量,可以预购买按小时计的 SQL DB 使用量,从而节省 SQL 计算成本。By using reserved capacity, you can pre-purchase SQL DB hourly usage and save over your SQL compute costs. SQL 许可证单独收费,不会因预留而打折。Your SQL license is charged separately and isn't discounted by the reservation. 预留容量是一项计费权益,它会自动应用于新部署和现有部署。Reserved capacity is a billing benefit and automatically applies to new and existing deployments. 顾问基于 3 年期预留定价并根据过去 30 天内观察到的使用模式进行推测,从而计算各订阅的成本节省估算值。Advisor calculates savings estimates for individual subscriptions by using 3-year reservation pricing and by extrapolating the usage patterns observed over the past 30 days. 共享范围建议适用于预留容量购买,并可提高成本节省。Shared scope recommendations are available for reserved capacity purchases and can increase savings. 有关详细信息,请参阅 AZURE Sql 数据库 & SQL 托管实例保留容量For details, see Azure SQL Database & SQL Managed Instance reserved capacity.

应用服务印花费预留容量App Service Stamp Fee reserved capacity

顾问会分析 Azure 应用服务隔离环境在过去 30 天内的印花费使用模式,并建议可优化成本的预留容量购买量。Advisor analyzes the Stamp Fee usage pattern for your Azure App Service isolated environment over the past 30 days and recommends reserved capacity purchases that optimize costs. 通过使用预留容量,可以为隔离环境印花费预购买按小时计的使用量,从而节省即用即付成本。By using reserved capacity, you can pre-purchase hourly usage for the isolated environment Stamp Fee and save over your pay-as-you-go costs. 请注意,预留容量仅适用于印花费,而不适用于应用服务实例。Note that reserved capacity applies only to the Stamp Fee and not to App Service instances. 预留容量是一项计费权益,它会自动应用于新部署和现有部署。Reserved capacity is a billing benefit and automatically applies to new and existing deployments. 顾问基于 3 年期预留定价并根据过去 30 天内的使用模式计算各订阅的成本节省估算值。Advisor calculates saving estimates for individual subscriptions by using 3-year reservation pricing based on usage patterns over the past 30 days.

Blob 存储预留容量Blob storage reserved capacity

顾问会分析 Azure Blob 存储和 Azure Data Lake Storage 在过去 30 天内的使用情况,Advisor analyzes your Azure Blob storage and Azure Data Lake storage usage over the past 30 days. 然后计算可优化成本的预留容量购买量。It then calculates reserved capacity purchases that optimize costs. 通过预留容量,可以预购买按小时计的使用量,从而节省按需成本。With reserved capacity, you can pre-purchase hourly usage and save over your current on-demand costs. Blob 存储预留容量仅适用于存储在 Azure Blob 常规用途 v2 和 Azure Data Lake Storage Gen2 帐户上的数据。Blob storage reserved capacity applies only to data stored on Azure Blob general-purpose v2 and Azure Data Lake Storage Gen2 accounts. 预留容量是一项计费权益,它会自动应用于新部署和现有部署。Reserved capacity is a billing benefit and automatically applies to new and existing deployments. 顾问基于 3 年期预留定价以及过去 30 天内观察到的使用模式计算各订阅的成本节省估算值。Advisor calculates savings estimates for individual subscriptions by using 3-year reservation pricing and the usage patterns observed over the past 30 days. 共享范围建议适用于预留容量购买,并可提高成本节省。Shared scope recommendations are available for reserved capacity purchases and can increase savings.

MariaDB、MySQL 和 PostgreSQL 预留容量MariaDB, MySQL, and PostgreSQL reserved capacity

顾问会分析 Azure Database for MariaDB、Azure Database for MySQL 和 Azure Database for PostgreSQL 在过去 30 天内的使用模式,Advisor analyzes your usage patterns for Azure Database for MariaDB, Azure Database for MySQL, and Azure Database for PostgreSQL over the past 30 days. 然后建议可优化成本的预留容量购买量。It then recommends reserved capacity purchases that optimize costs. 通过使用预留容量,可以预购买按小时计的 MariaDB、MySQL 和 PostgreSQL 使用量,从而节省当前成本。By using reserved capacity, you can pre-purchase MariaDB, MySQL, and PostgreSQL hourly usage and save over your current costs. 预留容量是一项计费权益,它会自动应用于新部署和现有部署。Reserved capacity is a billing benefit and automatically applies to new and existing deployments. 顾问基于 3 年期预留定价以及过去 30 天内观察到的使用模式计算各订阅的成本节省估算值。Advisor calculates savings estimates for individual subscriptions by using 3-year reservation pricing and the usage patterns observed over the past 30 days. 共享范围建议适用于预留容量购买,并可提高成本节省。Shared scope recommendations are available for reserved capacity purchases and can increase savings.

Azure Synapse Analytics(以前称为 SQL 数据仓库)预留容量Azure Synapse Analytics (formerly SQL Data Warehouse) reserved capacity

顾问会分析 Azure Synapse Analytics 在过去 30 天内的使用模式,并建议可优化成本的预留容量购买量。Advisor analyzes your Azure Synapse Analytics usage patterns over the past 30 days and recommends reserved capacity purchases that optimize costs. 通过使用预留容量,可以预购买按小时计的 Synapse Analytics 使用量,从而节省按需成本。By using reserved capacity, you can pre-purchase Synapse Analytics hourly usage and save over your on-demand costs. 预留容量是一项计费权益,它会自动应用于新部署和现有部署。Reserved capacity is a billing benefit and automatically applies to new and existing deployments. 顾问基于 3 年期预留定价以及过去 30 天内观察到的使用模式计算各订阅的成本节省估算值。Advisor calculates savings estimates for individual subscriptions by using 3-year reservation pricing and the usage patterns observed over the past 30 days. 共享范围建议适用于预留容量购买,并可提高成本节省。Shared scope recommendations are available for reserved capacity purchases and can increase savings.

删除未关联的公共 IP 地址可节省资金Delete unassociated public IP addresses to save money

顾问可识别与负载均衡器和 VM 等 Azure 资源无关的公共 IP 地址。Advisor identifies public IP addresses that aren't associated with Azure resources like load balancers and VMs. 这些公共 IP 地址会产生少许费用。A nominal charge is associated with these public IP addresses. 如果不打算使用它们,则可以通过删除它们来节省成本。If you don't plan to use them, you can save money by deleting them.

删除将要发生故障的 Azure 数据工厂管道Delete Azure Data Factory pipelines that are failing

顾问会检测重复失败的 Azure 数据工厂管道,Advisor detects Azure Data Factory pipelines that repeatedly fail. 并建议解决问题或在不需要时删除管道。It recommends that you resolve the problems or delete the pipelines if you don't need them. 即使这些管道在发生故障时没有为你提供服务,我们也会向你收取相关费用。You're billed for these pipelines even if though they're not serving you while they're failing.

使用托管磁盘的标准快照Use standard snapshots for managed disks

为了节省60% 的成本,我们建议将快照存储在标准存储中,而不考虑父磁盘的存储类型。To save 60% of cost, we recommend storing your snapshots in standard storage, regardless of the storage type of the parent disk. 此选项是托管磁盘快照的默认选项。This option is the default option for managed disk snapshots. 顾问标识存储在高级存储中的快照,并建议从高级存储迁移到标准存储。Advisor identifies snapshots that are stored in premium storage and recommends migrating then from premium to standard storage. 详细了解托管磁盘定价。Learn more about managed disk pricing.

使用生命周期管理Use lifecycle management

通过使用有关 Azure Blob 存储对象计数、总大小和事务的情报,顾问可检测是否应启用生命周期管理来对一个或多个存储帐户上的数据执行分层。By using intelligence about your Azure Blob storage object count, total size, and transactions, Advisor detects whether you should enable lifecycle management to tier data on one or more of your storage accounts. 它会提示你创建生命周期管理规则,以将数据自动分层到冷存储或存档存储,从而优化存储成本,并将数据保留在 Azure Blob 存储中来确保应用程序兼容性。It prompts you to create lifecycle management rules to automatically tier your data to cool or archive storage to optimize your storage costs while retaining your data in Azure Blob storage for application compatibility.

创建临时 OS 磁盘的建议Create an Ephemeral OS Disk recommendation

借助临时 OS 磁盘,你可以:Ephemeral OS Disk allows you to:

  • 节省 OS 磁盘的存储成本。Save on storage costs for OS disks.
  • 降低 OS 磁盘的读取/写入延迟。Get lower read/write latency to OS disks.
  • 通过将 OS(和临时磁盘)重置为其原始状态,加速 VM 重置映像操作。Get faster VM reimage operations by resetting the OS (and temporary disk) to its original state.

对于生存期较短的 IaaS VM 或具有无状态工作负荷的 VM,最好使用临时 OS 磁盘。It's preferable to use Ephemeral OS Disk for short-lived IaaS VMs or VMs with stateless workloads. 顾问提供有关可从临时 OS 磁盘中受益的资源的建议。Advisor provides recommendations for resources that can benefit from Ephemeral OS Disk.

缩短 Azure 数据资源管理器表缓存期(策略),实现群集成本优化(预览版)Reduce Azure Data Explorer table cache-period (policy) for cluster cost optimization (Preview)

顾问会识别符合以下要求的资源:缩短表缓存策略将释放具有低 CPU 利用率、内存和高缓存大小配置的 Azure 数据资源管理器群集节点。Advisor identifies resources where reducing the table cache policy will free up Azure Data Explorer cluster nodes having low CPU utilization, memory, and a high cache size configuration.

如何访问 Azure 顾问中的成本建议How to access cost recommendations in Azure Advisor

  1. 登录到 Azure 门户Sign in to the Azure portal.

  2. 在任意页面中搜索并选择顾问Search for and select Advisor from any page.

  3. 在顾问仪表板中,选择“成本”选项卡 。On the Advisor dashboard, select the Cost tab.

后续步骤Next steps

若要了解有关顾问建议的详细信息,请参阅以下资源:To learn more about Advisor recommendations, see: