雲端優先混合型架構的常見例子是,當含有重要資料的舊版應用程式和服務必須與新資料或應用程式整合時。如要完成整合,您可以使用混合式架構,透過 API 介面翻新舊版服務,讓舊版服務發揮效益,供新款雲端服務和應用程式使用。有了雲端 API 管理平台 (例如 Apigee),您就能在不大幅修改應用程式的情況下實作這類用途,並為舊版服務增加安全性、數據分析和擴充性。
遷移和現代化
混合式多雲端和 IT 現代化這兩種不同的概念彼此相關,並可形成良性循環。使用公用雲端可以促進及簡化 IT 工作負載的現代化。將 IT 工作負載現代化可讓您充分運用雲端服務。
舉例來說,您可以根據雲端式微服務架構,重構或重新設計大型的 VM 單體式應用程式,並將其轉換成數個獨立的微服務。在這個範例中,微服務架構會使用 Google Cloud 代管容器服務,例如 Google Kubernetes Engine (GKE) 或 Cloud Run。不過,如果目標雲端環境不支援應用程式的架構或基礎架構,您可以考慮先重新建構平台、重構或重新建構遷移策略,以便在可行情況下克服這些限制。
[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間:2025-01-23 (世界標準時間)。"],[[["\u003cp\u003eThis document guides users on migrating workloads to the cloud, focusing on hybrid and multicloud scenarios, as opposed to complete cloud migration.\u003c/p\u003e\n"],["\u003cp\u003eA cloud-first approach involves deploying new workloads to the public cloud, potentially limiting the benefits and increasing IT complexity, and thus is best for selected workloads.\u003c/p\u003e\n"],["\u003cp\u003eModernizing IT workloads in conjunction with cloud migration can enhance agility, reduce costs, and improve reliability, however, not every application may be fit for modernization in the initial migration phase.\u003c/p\u003e\n"],["\u003cp\u003eMultiple migration types such as rehosting, replatforming, refactoring, rearchitecting, rebuilding, and repurchasing, can be combined and applied to different workloads based on business and technical needs.\u003c/p\u003e\n"],["\u003cp\u003eWorkload selection criteria should be applied to determine if a rehost, refactor, or rebuild migration strategy is most applicable, based on factors such as environment dependencies, compute resource usage, and third-party application requirements.\u003c/p\u003e\n"]]],[],null,["# Architectural approaches to adopt a hybrid or multicloud architecture\n\nThis document provides guidance on common and\nproven approaches and considerations to migrate your workload to the cloud. It\nexpands on guidance in\n[Design a hybrid and multicloud architecture strategy](/architecture/hybrid-multicloud-patterns/strategy#design_a_hybrid_and_multicloud_architecture_strategy),\nwhich discusses several possible, and recommended, steps to design a strategy\nfor adopting a hybrid or multicloud architecture.\n| **Note:** The phrase *migrate your workload to the cloud* refers to hybrid and multicloud scenarios, not to a complete cloud migration.\n\nCloud first\n-----------\n\nA common way to begin using the public cloud is the *cloud-first* approach.\nIn this approach, you deploy your new workloads to the public cloud while your\nexisting workloads stay where they are. In that case, consider a classic\ndeployment to a private computing environment only if a public cloud deployment\nis impossible for technical or organizational reasons.\n\nThe cloud-first strategy has advantages and disadvantages. On the positive side,\nit's forward looking. You can deploy new workloads in a modernized fashion while\navoiding (or at least minimizing) the hassles of migrating existing workloads.\n\nWhile a *cloud-first* approach can provide certain advantages, it could\npotentially result in missed opportunities for improving or using existing\nworkloads. New workloads might represent a fraction of the overall IT landscape,\nand their effect on IT expenses and performance can be limited. Allocating time\nand resources to migrating an existing workload could potentially lead to more\nsubstantial benefits or cost savings compared to attempting to accommodate a new\nworkload in the cloud environment.\n\nFollowing a strict *cloud-first* approach also risks increasing the overall\ncomplexity of your IT environment. This approach might create redundancies,\nlower performance due to potential excessive cross-environment communication, or\nresult in a computing environment that isn't well suited for the individual\nworkload. Also, compliance with industry regulations and data privacy laws can\nrestrict enterprises from migrating certain applications that hold sensitive\ndata.\n\nConsidering these risks, you might be better off using a cloud-first approach\nonly for selected workloads. Using a cloud-first approach lets you concentrate\non the workloads that can benefit the most from a cloud deployment or migration.\nThis approach also considers the modernization of existing workloads.\n\nA common example of a cloud-first hybrid architecture is when legacy\napplications and services holding critical data must be integrated with new data\nor applications. To complete the integration, you can use a hybrid\narchitecture that modernizes legacy services by using API interfaces, which\nunlocks them for consumption by new cloud services and applications.\nWith a cloud\n[API management platform](/solutions/unlocking-legacy-applications),\nlike\n[Apigee](/apigee),\nyou can implement such use cases with minimal application changes and add\nsecurity, analytics, and scalability to the legacy services.\n\nMigration and modernization\n---------------------------\n\nHybrid multicloud and IT modernization are distinct concepts that are linked in\na virtuous circle. Using the public cloud can facilitate and simplify the\nmodernization of IT workloads. Modernizing your IT workloads can help you\nget more from the cloud.\n\nThe primary goals of modernizing workloads are as follows:\n\n- Achieve greater agility so that you can adapt to changing requirements.\n- Reduce the costs of your infrastructure and operations.\n- Increase reliability and resiliency to minimize risk.\n\nHowever, it might not be feasible to modernize every application in the\nmigration process at the same time. As described in\n[Migration to Google Cloud](/architecture/migration-to-gcp-getting-started),\nyou can implement one of the following\n[migration types](/architecture/migration-to-gcp-getting-started#types_of_migrations),\nor even combine multiple types as needed:\n\n- Rehost (lift and shift)\n- Replatform (lift and optimize)\n- Refactor (move and improve)\n- Rearchitect (continue to modernize)\n- Rebuild (remove and replace, sometimes called *rip and replace*)\n- Repurchase\n\nWhen making strategic decisions about your hybrid and multicloud architectures,\nit's important to consider the feasibility of your strategy from a cost and time\nperspective. You might want to consider a phased migration approach, starting\nwith lifting and shifting or replatforming and then refactoring or\nrearchitecting as the next step. Typically, lifting and shifting helps to\noptimize applications from an\ninfrastructure perspective. After applications are running in the cloud, it's\neasier to use and integrate cloud services to further optimize them using\ncloud-first architectures and capabilities. Also, these applications can still\ncommunicate with other environments over a hybrid network connection.\n\nFor example, you can refactor or rearchitect a large, monolithic VM-based\napplication and turn it into several independent microservices, based on a\ncloud-based microservice architecture. In this example, the microservices\narchitecture uses Google Cloud managed container services like\n[Google Kubernetes Engine (GKE)](/kubernetes-engine)\nor\n[Cloud Run](/run/docs/overview/what-is-cloud-run).\nHowever, if the architecture or infrastructure of an application isn't supported\nin the target cloud environment as it is, you might consider starting with\nreplatforming, refactoring, or rearchitecting your migration strategy to\novercome those constraints where feasible.\n\nWhen using any of these migration approaches, consider modernizing your\napplications (where applicable and feasible). Modernization can require adopting\nand implementing\n[Site Reliability Engineering (SRE)](/sre)\nor DevOps principles, such that you might also need to extend application\nmodernization to your private environment in a hybrid setup. Even though\nimplementing SRE principles involves engineering at its core, it's more of a\ntransformation process than a technical challenge. As such, it will likely\nrequire procedural and cultural changes. To learn more about how the first step\nto implementing SRE in an organization is to get leadership buy-in, see\n[With SRE, failing to plan is planning to fail](https://cloud.google.com/blog/products/devops-sre/sre-success-starts-with-getting-leadership-on-board).\n\nMix and match migration approaches\n----------------------------------\n\nEach migration approach discussed here has certain strengths and weaknesses. A\nkey advantage of following a hybrid and multicloud strategy is that it isn't\nnecessary to settle on a single approach. Instead, you can decide which approach\nworks best for each workload or application stack, as shown in the following\ndiagram.\n\nThis conceptual diagram illustrates the various migration and modernization\npaths or approaches that can be simultaneously applied to different workloads,\ndriven by the unique business, technical requirements, and objectives of each\nworkload or application.\n\nIn addition, it's not necessary that the same application stack components\nfollow the same migration approach or strategy. For example:\n\n- The backend on-premises database of an application can be replatformed from self-hosted MySQL to a managed database using [Cloud SQL](/sql) in Google Cloud.\n- The application frontend virtual machines can be refactored to run on containers using [GKE Autopilot](/kubernetes-engine/docs/concepts/autopilot-overview), where Google manages the cluster configuration, including nodes, scaling, security, and other preconfigured settings.\n- The on-premises hardware load balancing solution and web application firewall WAF capabilities can be replaced with [Cloud Load Balancing](/load-balancing) and [Google Cloud Armor](/armor).\n\nChoose rehost (lift and shift), if any of the following is true of the\nworkloads:\n\n- They have a relatively small number of dependencies on their environment.\n- They aren't considered worth refactoring, or refactoring before migration isn't feasible.\n- They are based on third-party software.\n\nConsider refactor (move and improve) for these types of workloads:\n\n- They have dependencies that must be untangled.\n- They rely on operating systems, hardware, or database systems that can't be accommodated in the cloud.\n- They aren't making efficient use of compute or storage resources.\n- They can't be deployed in an automated fashion without some effort.\n\nConsider whether rebuild (remove and replace) meets your needs for these\ntypes of workloads:\n\n- They no longer satisfy current requirements.\n- They can be incorporated with other applications that provide similar capabilities without compromising business requirements.\n- They are based on third-party technology that has reached its end of life.\n- They require third-party license fees that are no longer economical.\n\nThe\n[Rapid Migration Program](/solutions/cloud-migration-program)\nshows how Google Cloud helps customers to use best practices, lower risk,\ncontrol costs, and simplify their path to cloud success."]]