There are many motivations for evolving from an entirely on-prem infrastructure to a multiple or hybrid cloud architecture. From the very beginning of the cloud adoption process, hybrid cloud architectures allow enterprises to benefit from cloud economics and scalability without compromising data sovereignty. A multicloud storage deployment also brings many benefits to the enterprise cloud, from avoiding vendor lock-in to accommodating mergers and acquisitions and optimizing price/performance.
In this blog post, we’ll look at everything related to multicloud/hybrid architectures: the use cases best served by them, the challenges of hybrid and multicloud management, how Cloud Volumes ONTAP helps to make both options easy and cost-effective, including customers case studies.
Multicloud and Hybrid Architectures: A Primer
Both multicloud and hybrid are distributed deployment models in the cloud, for running an app—or part of an app—in the most appropriate computing environment.
What is Multicloud?
Multicloud is the use of cloud computing and storage services—IaaS, PaaS, SaaS—from two or more public cloud vendors in a single network architecture. A multicloud strategy avoids vendor lock-in, enhances business continuity, and maximizes flexibility for development teams. However, multicloud management can be complicated, requiring careful orchestration as well as close attention to security and governance issues.
What is Hybrid Cloud?
A hybrid cloud combines a privately-hosted cloud with at least one public cloud service provider, all managed as a single, policy-based environment. In a hybrid enterprise cloud deployment, on-prem infrastructure is extended by public cloud resources in a wide range of use cases such as dev/test, backup and archiving, bursting production workloads to meet sudden demand, and more. A common reason for using a hybrid environment is as a transition to a full cloud deployment.
Both major cloud providers are catering to this new demand. For AWS, hybrid cloud deployment can use AWS Storage Gateway and many of its other services. Not to be left out of the shift to the hybrid cloud, Azure also offers a solution: StorSimple. In addition to those native hybrid solutions, many AWS and Azure services can also be used in hybrid deployments with the help of Cloud Volumes ONTAP, which we’ll discuss later.
Hybrid and Multicloud Deployment Models
When it comes to either deployment model in the cloud, whether hybrid or multicloud, there are two main types: distributed and redundant.
In a distributed architecture some apps or app components will run in one environment (e.g., on-prem or on a particular public cloud) while other apps or other components of the same app will run in another environment. Some examples of distributed deployment and data storage include:
- Tiered hybrid: For example, the frontend runs in the public cloud for greater agility and performance while the backend, including data storage, runs in the private cloud.
- Partitioned multicloud: The enterprise maintains multiple cloud environments and deploys an app and its data where most suitable based on, for example, geographic location of availability zones or specialized services provided by a particular cloud provider.
- Analytics hybrid/multicloud: Transactional workloads run close to home in a private cloud, but the data collected by the workloads is extracted and loaded into a public cloud for analytics.
In redundant architectures, the same app or app components will be deployed across multiple computing and storage environments in order to increase capacity or resiliency. This kind of architecture is often adopted by companies just starting to use the cloud, as it allows them to gain experience with the platform before they go all in. In a redundant architecture, there is usually no need to refactor or rewrite code and processes. Instead they can be lifted and shifted “as is” to the cloud infrastructure. Some examples of redundant deployment include:
- Environment hybrid: Production workloads are run in the private cloud, but replicated workloads for non-production purposes such as development, testing, staging are run in the public cloud.
- Business continuity multicloud or hybrid cloud storage: Backups, archives, DR sites, standby systems are deployed redundantly across multiple clouds in order to avoid single point of failure (SPOF) scenarios.
- Cloud bursting: Baseline workloads run privately, but they can burst to the cloud as needed in order to scalably meet elastic capacity demands.
According to RightScale’s 2018 State of the Cloud Report, over 80% of North American and European companies are using a complex deployment model in the cloud: 51% of them hybrid and 21% implementing a multicloud strategy, with an average of 5 cloud providers.