Multi-Cloud: A Complete Guide

Data is a strategic resource for any organization. From providing insights about customer behavior to determining the direction of product development, this information provides the detail for making informed decisions.

But too often, company data has untapped potential. It might be siloed away in on-premises infrastructure, locked into the ecosystem of a single cloud provider, or constrained by the performance limitations of a cloud-only data warehouse.

A multi-cloud architecture provides a way out, delivering the choice, savings, and shortened path to insight that companies need as data volumes expand and workload complexity grows. 

The benefits of a multi-cloud platform

What is multi-cloud?

Multi-cloud is the running of applications in multiple clouds. Multiple cloud providers like AWS, Google Cloud, and Microsoft Azure deliver the computing, cloud storage, and network resources to the customer in a multi-cloud deployment.

This approach lets organizations choose which services from which cloud service provider (CSP) to consume, so they’re never 100% reliant on one CSP.

For example, a multi-cloud setup could feature a production environment running in Public Cloud A, development and testing in Public Cloud B, and disaster recovery (DR) in Public Cloud C. Likewise, data analytics may be pulled from different data pipelines associated with multiple public cloud providers and processed by a connected cloud data platform.

Overall, a multi-cloud strategy lets organizations connect data from any cloud or on-premises source and accelerate their analytics initiatives. This is important as enterprises adopt more Software-as-a-Service (SaaS) applications and look to access their analytics in real time.

Multi-cloud and the new shape of data analytics

Multi-cloud has become one of the most popular types of cloud deployment, to the extent that it is redefining the possibilities of a data analytics strategy. Gartner estimated that 81% of enterprises were using at least two public clouds in 2020, for two overarching reasons:

  1. Avoiding vendor lock-in: A company doesn’t have to depend exclusively on a CSP’s roadmap or technologies, nor risk losing access if that CSP experiences an outage.

  2. Choosing best-of-breed solutions: Organizations can pick services from any cloud vendor, based on criteria like functionality, price, and geographic location.

For analytics workloads in particular, these characteristics of a multi-cloud strategy give enterprises the flexibility to run any query on the cloud infrastructure(s) that will deliver the best combination of cost and performance. For example, Dutch chain store Intertoys pursued a multi-cloud migration strategy to modernize the company’s analytics and run complex queries:

  • Data volumes were growing and coming in from an expanding variety of sources. This is a widespread challenge for enterprises; a CIO survey found that organizations averaged 400 data sources and that 20% of companies relied on 1,000 or more.

  • Multiple data schema were also in play, creating complicated mixed workloads. Structured, semi-structured, and unstructured data had to be processed to support everything from simple reports to executive dashboards.

  • Complex multi-join queries demanded the highest level of performance and multidimensional scalability, at the most economical cost. This combination is only possible with a multi-cloud deployment that keeps every option open.

Intertoys deployed a connected cloud data platform on the Microsoft Azure public cloud. This migration ensured the company got the elasticity and cost optimization to extract the most value from hybrid multi-cloud data workloads.

What are the biggest benefits of multi-cloud?

No lock-in and expanded choice are the two core benefits of using multiple cloud platforms, underpinning all of the advantages of this deployment model. Others include:

Analytics anywhere

Migration to a multi-cloud environment enables organizations to pick the best cloud and toolset for them for each analytics workload. A company can match cloud compute and storage solutions to its particular data sources and business requirements, optimizing cost and performance. Any cloud becomes a potential option for any workload, and analytics aren’t beholden to a single CSP.

Accelerated innovation

Enterprises can freely shop for the latest cloud computing services that will accelerate their analytics. They don’t have to wait on Cloud Vendor A if Cloud Vendor B has already innovated a service that fits their needs. The expanded choice available via multi-cloud also simplifies the creation of cloud environments for specific use cases, such as better management of healthcare data.

Improved resiliency and redundancy

Relying exclusively on a one-cloud stack puts enterprises at high risk from an outage — the “all eggs in one basket” problem. But multi-cloud helps them diversify. They can use a cloud for backup — a different cloud than the one used for production workloads — enabling data to be failed-over and moved between clouds. Or they might add a cloud DR service to their on-premises environment. Multi-cloud enables more resilient performance.

Cost optimization

Going multi-cloud may lower cloud costs, because organizations can switch between services based on changes in price. They can also search for the combination of compute and storage that will support the most cost-effective workload scaling.

How do you build a multi-cloud deployment?

Moving to a multi-cloud deployment is a multi-step process:

  1. Define the scope of the project. Multi-cloud migration could mean moving all applications into a public cloud, or just a subset of them, with the remainder still on-premises.

  2. Perform workload placement analysis, to see where each makes the most sense from a performance and cost perspective. Think about if, and how, on-premises resources fit in.

  3. Consider CSPs. Each cloud vendor may excel in a particular area, such as its pricing, performance, or tooling. The provider’s cybersecurity practices and service-level agreements are important factors, too.

  4. Evaluate data platform options. Because avoiding data silos and data drift is a core multi-cloud challenge, the right multi-cloud solution will be able to ingest data from any source and deliver consistent, economical, and scalable performance.

During this process, it may become apparent that a hybrid cloud consisting of one cloud computing service and an on-premises environment is a better option than a multi-cloud route. Alternatively, a hybrid multi-cloud setup could be the right choice.

Hybrid cloud vs. multi-cloud, how do they compare?

Different companies have different definitions of hybrid cloud.

Teradata’s view is that hybrid cloud always involves an on-premises infrastructure, while multi-cloud can (as part of a hybrid multi-cloud deployment), but doesn’t have to include this on-premises component in practice. Multi-cloud often means just multiple public cloud services.

Other definitions of hybrid cloud may define it as a combination of public cloud and private cloud resources, with private cloud defined as a cloud environment dedicated to one customer.

Organizations do not have to choose between the two, as they can be combined. There are distinctive advantages and disadvantages to each on its own, though:

Multi-cloud architecture


  • Choose any service from any public cloud provider.

  • Spread data and risk across multiple public clouds.

  • Avoid vendor lock-in and shop for the best prices.


  • Increased complexity, possibly requiring big changes to data architecture.

  • Security is a heightened concern due to a larger attack surface.

  • Some vendor-specific discounts might not be available.

Hybrid cloud architecture


  • Preserve on-premises data and infrastructure investments.

  • Maintain tight control over security and performance.

  • Use on-demand resources economically, only as needed.


  • On-premises operations and maintenance can be time-consuming.

  • Must ensure API-consistent and secure connections between environments.

  • Implementation is difficult without the right data platform and tooling.

How secure are multi-cloud and hybrid cloud environments?

A multi-cloud environment is only as secure as the cybersecurity practices of its cloud providers as well as the customer's users themselves. Multi-cloud strategies can increase security complexity, because data is living in more places, but major CSPs are on the cutting edge of cybersecurity and have implemented multi-layered security for their services.

The exact extent of this CSP-provided security varies by the type of service. Software-as-a-Service security is almost entirely the responsibility of the cloud provider, while Infrastructure-as-a-Service and Platform-as-a-Service are more of a shared responsibility between the provider and customer.

Hybrid cloud introduces additional security considerations because the customer organization continues to manage its on-premises resources. Tasks like patching, updating, and managing the lifecycle of hardware and software remain incumbent on its internal teams.

Moving forward with multi-cloud for analytics

A multi-cloud environment works best with a connected cloud data platform that supports consistent performance, multidimensional scalability, and a single source of truth. Such a solution can connect to and synchronize anything, and address any need from basic data warehousing to advanced analytics.

Most importantly, it future-proofs analytics. Organizations get to keep their cloud service options open, while being able to bank on their data platform to work seamlessly with any CSP or solution.

Teradata Vantage is the connected multi-cloud data platform that delivers consistent performance and cost-effectiveness for data workloads of any type. Learn more about how we can help you enable a connected multi-cloud by contacting our team.