您可以选择自行托管 Looker（原始版本）实例或部署，也可以让 Looker 为您托管。在我们的文档中，我们将由 Looker 托管的实例或部署称为“Looker 托管”，将本地托管的实例或部署称为“客户托管”或“自行托管”。
如果 Looker 托管您的部署，Looker 会根据资源利用率和业务需求，代表您管理与 Looker 应用相关的所有必要 IT 功能，从而大大减少安装、配置和维护 Looker 应用所需的工作量。相反，如果您自行托管部署，则需要负责管理许多此类流程和功能。客户托管部署包含产品内服务，即由 Looker 托管并通过产品访问的服务，具体包括许可数据、配置备份、系统错误报告、数据操作和支持工单，如 Looker 共享的应用数据 部分中所述，该部分位于 Looker 的安全页面。
在两种托管选项之间进行选择，需要在便捷性和控制力之间做出权衡。选择 Looker 托管可让您专注于将 Looker 集成到业务工作负载中，而无需进行基础设施管理。相反，自行托管 Looker 部署可让您完全控制基础设施管理，但会增加初始启动和持续维护的开销。
本页提供了以下信息，以帮助您选择最适合您需求的托管选项：
每种托管选项的比较优势
下表比较了每种托管选项的优势。
| 优势 | Looker 托管 | 客户托管 |
|
默认访问所有 Looker 功能 |
X |
|
|
无需进行硬件设置或维护 |
X |
|
|
自动软件更新和维护（每月或每季度） |
X |
|
|
免费扩缩硬件：
|
X |
|
|
应用和主机监控 |
X |
|
|
自动 Looker 实例备份 |
X |
|
|
后端数据库迁移（如有必要） |
X |
|
|
正常运行时间 SLA 99.9%（高级版和精英版） |
X |
|
|
S1 响应 SLA 1 小时 |
X |
|
|
Looker 状态信息中心 |
X |
|
|
使用 Looker API |
X |
X |
|
多实例迁移 |
X |
X |
|
直接访问和导出服务器及日志 |
X |
Looker 托管选项的优势和限制
下表列出了使用 Looker 托管部署的优势和限制。
Looker 托管部署的优势
下表列出了使用 Looker 托管部署的优势。
|
主动管理 Looker 实例 |
将性能监控工作交给 Looker 团队，专注于利用 Looker 提供的可操作数据分析来制定决策。 |
|
始终获取最新功能和更新 |
您无需手动下载其他更新。Looker 会测试、更新和优化您的 Looker 实例。 |
|
性能一致 |
Looker 会监控性能并根据需要调整容量。您无需决定需要多少服务器来支持用户。 |
|
部署安全性 |
Looker 会管理您的平台基础设施。您的 Looker 部署是离散的、安全的，并且会持续受到监控，以确保您的数据安全。 |
|
关联服务 |
每个 Looker 部署都包含必要的关联服务，可让您访问更多相关数据、深入分析数据、简化和扩缩数据建模，并与第三方系统集成，同时 Looker 还会提供支持。 |
|
安全的数据库连接 |
使用 SSH 将 Looker 连接到您的数据库。 |
|
SAML/LDAP 集成 |
Looker 可以与您现有的身份验证方法集成。 |
|
实例状态监控 |
您可以通过 Looker 状态信息中心 了解实例在产品中断/服务暂停方面的实时动态信息。 |
Looker 托管部署的限制
在选择使用 Looker 托管部署之前，请考虑下表中列出的限制。
|
特定的安全/合规要求 |
Looker 托管环境基础设施可能与 贵公司的个人安全/合规要求不一致。 |
|
需要能够导出日志/监控信息 |
由于 Looker 会管理实例的基础设施，因此您无法导出实例使用情况的日志。Looker 会管理所有监控。 |
|
自定义 SLA |
Looker 更新和维护会在预定义的 维护窗口期间进行。部分客户可能需要对维护发生的时间进行额外控制。 |
|
自定义 JDBC 驱动程序 |
|
|
查看底层托管基础设施详情 |
由于 Looker（原始版本）是一个全代管式环境，因此底层托管基础设施（例如 Google Cloud 项目或 AWS 环境）由 Google 在内部管理，并且不会向客户公开详细信息。Looker 界面中没有自助式管理员配置面板，客户拥有的云控制台中也无法查看这些详细信息。 |
客户托管选项的优势和限制
下表列出了使用客户托管部署的优势和限制。
客户托管部署的优势
下表列出了使用客户托管部署的优势。
|
直接控制基础设施和伸缩决策 |
您可以实现 Looker 托管部署可能不提供的基础设施和架构配置。 |
|
访问日志记录/监控 |
通过管理自己的基础设施，您可以直接访问和导出 Looker 应用日志，并设置符合您个人要求的实例监控。 |
|
定制的安全模型 |
在自己的部署中托管 Looker 可让您完全控制 应用环境的安全性，您可以根据公司/行业特定的安全标准进行调整。 |
客户托管部署的限制
在选择使用客户托管部署之前，请考虑下表中列出的限制 。
|
支持限制 |
如果 Looker 支持团队不熟悉自定义部署架构，则问题排查可能会很困难。某些问题可能需要贵组织更多参与。 |
|
每月更新要求 |
您需要负责创建和维护流程，以确保用户从 Looker 获取最新功能和安全补丁。 |
|
人力资本要求 |
Looker 部署可能需要大量人员和站点可靠性工程专业知识。贵组织必须管理 Looker 部署的各个组件。 |
|
费用 |
贵组织必须管理时间、人力资本和云/数据中心费用。 |
|
使用关联服务时遇到的挑战 |
Looker 具有关联服务的优势，可让您访问相关行业数据、深入分析数据、简化和扩缩数据建模，并与第三方系统集成。在自行托管的 Looker 部署中，您可能需要在云中部署辅助服务才能访问这些功能。 |
|
灾难恢复和弹性 |
您需要负责维护正常运行时间和服务的弹性。 |
|
部分 Looker 功能不可用 |
如需比较 Looker 的各种托管选项对功能的支持情况，请参阅本页面的 Looker 功能支持（按托管选项） 部分。 |
|
其他功能配置 |
客户托管实例可能需要额外配置才能使用某些 Looker 功能。 |
客户托管部署的使用情形示例
如果客户托管选项符合以下任何使用情形，则可能适合您的部署：
- 贵组织有定制的安全要求：部分组织的安全政策规定，他们无法使用云服务。Looker（原始版本）旨在实现多云，Looker 托管的 Looker（原始版本）部署位于各种云服务提供商（例如 Google CloudAWS 和 Azure）上。如果您的安全政策与使用云服务处理数据不兼容，则客户托管旨在成为替代方案。
- 贵组织偏好完全可自定义的部署模型：如果 Looker 托管您的环境，Looker 会假定这些环境旨在保持稳定。贵组织的要求可能与此假设不兼容，例如在以下情况下：
- 您的部署环境是针对每个用户群组或客户的大量实例，这些实例可能仅在短暂的时间内需要。
- 贵组织需要能够频繁构建新环境或拆除现有环境。
- 贵组织需要多个 Looker 实例，每个实例都需要针对启动标志、模型或连接信息进行自定义配置。
- 贵组织需要更直接地访问集成和配置功能：在 Looker 托管的环境中，您对部署的直接访问受到限制。客户托管部署允许完全访问实例的文件系统、元数据数据库和 JVM 配置。在以下情况下，直接访问可能很有用：
- 您经常使用与开发流程同步的脚本更新每个实例的 LookML 模型和配置。
- 贵组织无法使用 Looker 用于部署的某些核心技术，例如 Git。通过完全控制实例的后端，您可以将 Looker 的任何核心组件替换为您首选的解决方案。
Looker 功能支持（按托管选项）
下表比较了 Looker 托管的 Looker（原始版本）、Looker (Google Cloud Core) 和客户托管的 Looker（原始版本）实例类型的功能支持情况。
Filter the table on supported instance types:
| Feature | Looker instance type | |||
|---|---|---|---|---|
| Platform features | Looker (original) | Looker (Google Cloud core) | Customer-hosted Looker (original) | Notes |
| Looker reports: Create and access reports in Looker | Yes | Yes | No | Not available for Looker (Google Cloud core) instances that use VPC Service Controls or CMEK. Not all Looker reports features are available on Looker (Google Cloud core) instances. See the Looker reports documentation for details. |
| Gemini in Looker: Conversational Analytics, Code Interpreter, LookML writing assistance, and Visualization Assistant | Yes | Yes | No | Gemini assistance is not supported for Looker (Google Cloud core) instances that use CMEK or VPC Service Controls or for Looker (Google Cloud core) instances that are connected to Data Studio through the Looker connector. |
| Looker-managed MCP server: A Model Context Protocol (MCP) server provided directly by the Looker platform. | Yes | Yes | No |
The Looker-managed MCP server is in preview. It is the recommended method for connecting AI applications to Looker instances that are hosted by Looker. Customer-hosted instances can use the standalone MCP Toolbox instead. |
| BI Connectors: Connected Sheets and Looker Studio: Use Google Sheets to explore data from your LookML models or connect to Looker Studio data in your Looker instance. | Yes | Yes | No |
Complimentary licenses to use Looker Studio Pro are not available for customer-hosted Looker (original) instances. For private connections configurations in Looker (Google Cloud core) instances, your network administrator must configure your network so that traffic from Google Sheets or Looker Studio can reach your Looker instance. |
| BI Connectors: Microsoft Excel, Microsoft Power BI, Tableau Desktop: Data connectors to Excel, Power BI, and Tableau. | Yes | Yes | Yes | For the Power BI or Tableau connectors for Looker (Google Cloud core) instances that use private connections and for customer-hosted Looker (original) instances, your network administrator must configure your network so that traffic from Power BI or Tableau can reach your Looker instance. |
| Looker Action Hub and Looker Marketplace | Yes | Yes | Yes |
This feature may have limitations for Looker (Google Cloud core) instances that use private connections. Additional configuration may also be necessary for Looker (Google Cloud core) instances that use private connections. Additional configuration may be necessary for customer-hosted Looker (original) instances. |
| Schedule data to S3, SFTP, and webhook | Yes | Yes | Yes | This feature may have limitations for Looker (Google Cloud core) instances that use private connections. Additional configuration may also be necessary for Looker (Google Cloud core) instances that use private connections. |
| Looker extension for VS Code: Local LookML development in VS Code-based IDEs. | Yes | Yes | Yes | Supports syntax highlighting, autocomplete, real-time file synchronization, and LookML validation. Currently in preview. |
| Looker mobile application | Yes | Yes | Yes | |
| Custom themes | Yes | Yes | Yes | Custom themes are available only for Enterprise and Embed editions of Looker (Google Cloud core). |
| Custom extensions | Yes | Yes | Yes | Custom extensions are available only for Enterprise and Embed editions of Looker (Google Cloud core). |
| Knowledge Catalog integration: Automatically ingest Looker metadata into Knowledge Catalog for discovery | No | Yes | No | |
| Knowledge Catalog data lineage: Trace data flow from BigQuery to Looker dashboards | No | Yes | No | Supported only for BigQuery data sources. |
| Private embedding | Yes | Yes | Yes |
Private embedding is available for all editions of Looker (Google Cloud core). |
| Signed embedding | Yes | Yes | Yes |
Signed embedding is available only for the Embed edition of Looker (Google Cloud core). |
| Preview features | Yes | Yes | Yes | Looker preview features are enabled on Looker (Google Cloud core). However, preview features that are categorized as Labs features are not available in Looker (Google Cloud core). |
| Legacy features | Yes | No | Yes | |
| Security and authentication features | Looker (original) | Looker (Google Cloud core) | Customer-hosted Looker (original) | Notes |
| Security features: private connections, CMEK, VPC Service Controls | No | Yes | No | Private connections and VPC Service Controls are available only for Enterprise and Embed editions of Looker (Google Cloud core). |
| IAM integration | No | Yes | No | |
| SAML authentication, OpenID authentication | Yes | Yes | Yes | |
| Username and password setting for login, LDAP authentication | Yes | No | Yes | |
| IP Allowlist setting | Yes | Yes | No | |
| Ability to sudo as another user | Yes | No | Yes | For the Embed edition of Looker (Google Cloud core), impersonation of embed users is available by using the API endpoint Login user. |
| FIPS encryption | No | Yes | Yes | FIPS encryption is available only in the Enterprise and Embed editions of Looker (Google Cloud core). |
| Compliance certifications (for example, FedRAMP High and other Assured Workloads products) | No | Yes | Yes | Compliance certifications are available only in the Enterprise and Embed editions of Looker (Google Cloud core). |
| LookML development and database connections | Looker (original) | Looker (Google Cloud core) | Customer-hosted Looker (original) | Notes |
| Dialect support | Yes | Yes | Yes | The documentation lists the dialects that are supported by Looker (Google Cloud core) and the dialects that are supported by Looker (original) and customer-hosted Looker (original). |
| Customer-hosted Git on the public internet | Yes | Yes | Yes | This feature may have limitations for Looker (Google Cloud core) instances that use private connections. Additional configuration may also be necessary for Looker (Google Cloud core) instances that use private connections. |
| Looker Continuous Integration | Yes | Yes | No | Not available for Looker (Google Cloud core) instances that use private connections or CMEK. |
| Service agent access to BigQuery | No | Yes | No | |
| SQL Runner visualizations: Create visualizations directly in SQL Runner. | Yes | No | Yes | SQL Runner Vis is a Labs feature and is not available in Looker (Google Cloud core). |
| SSH tunnels (for on-premises databases) | Yes | n/a | Yes | |
| Logging | Looker (original) | Looker (Google Cloud core) | Customer-hosted Looker (original) | Notes |
| Access to internal database and verbose logs | No (see note) | No | Yes | For Looker-hosted Looker (original) instances, internal database and verbose logs are available only through a Google Cloud support request. |
| Cloud Audit Logs | No | Yes | No | |
| Elite System Activity | Yes | Yes | No |
Elite System Activity is available only as a paid feature in the Elite version of Looker (original). Elite Style Activity is available only in the Enterprise and Embed editions of Looker (Google Cloud core). |
| Infrastructure and instance management | Looker (original) | Looker (Google Cloud core) | Customer-hosted Looker (original) | Notes |
| Fully Google Cloud-managed Google service | No | Yes | No | Google Cloud support has a much greater ability to troubleshoot issues with the Looker (Google Cloud core) environment, because it is managed on Google Cloud. |
| Self-service instance creation | No | Yes | No | |
| Terraform support | No | Yes | No | |
| Startup options | Yes | No | Yes | |
| Customer-defined maintenance windows | No | Yes | No | |
| Non-production instances | Yes | Yes | Yes |
Two non-production instances are available for the Elite version of Looker (original). No other Looker (original) versions offer non-production instances. Non-production instances are purchased separately for Looker (Google Cloud core). |
| Export | Yes | No | No |
This feature is required to migrate from Looker (original) to Looker (Google Cloud core). |
| Release channels | No | Yes | No | This feature is in preview. Looker (original) doesn't have defined release channels, but does have an extended support release (ESR) program in addition to its standard release program. |