As companies embrace cloud-native architectures and scale their operations, ensuring business continuity and minimizing downtime become paramount; even brief outages can result in substantial revenue loss and reputational damage.
In July 2023, a flaw in CrowdStrike's software update caused widespread disruption across various industries. The air travel sector was particularly hard hit, with over 3,300 flights cancelled globally. Major airlines like United, Delta, American Airlines, British Airway. The ripple effects extended beyond air travel, impacting payment systems, banking services, healthcare providers, and even emergency services in some regions.
While the exact financial impact of this outage is still being calculated at the time of writing, it's clear that the reliance on a single cloud provider or a monolithic IT infrastructure is increasingly becoming a risky strategy. Medium to large corporations with critical data and operations could potentially save millions in the long run by implementing a robust disaster recovery strategy, particularly one that leverages multi-cloud architecture. Therefore, this blog will explore multi-cloud, its promises, and why you should care.
An introduction to Multi-Cloud
At its core, Multi-Cloud is a deployment model that leverages two or more public clouds. This is not to be confused with a hybrid deployment model that leverages a public cloud provider in addition to on-premise or private data centers.
This video from HashiCorp highlights what Multi-Cloud is and why companies are adopting it:
Multi-Cloud vs. Single Cloud at a glance
What are the challenges of Single-Cloud?
Recent events have underscored just how catastrophic a single cloud provider's failure can be. Take the case of Australian pension fund UniSuper, which in May 2024 had its entire Google Cloud account containing $135 billion in funds and data for 647,000 members completely wiped out - including backups.
If UniSuper relied solely on Google Cloud, this would have been disastrous. Thankfully, they also maintained backups with an alternative provider, allowing them to eventually recover. However, the two-week downtime highlights the severe impact of vendor outages.
According to Mark Boost, the CEO of Civo, this incident vividly demonstrates the inherent risks of a single-cloud strategy. He emphasizes that the fragility of concentrating power with a few tech giants can lead to significant vulnerabilities. He advocates for a multi-cloud approach, where organizations utilize multiple cloud providers to mitigate risks, enhance resilience, and ensure continuity of services. This strategy not only protects against technical failures but also shields businesses from monopolistic practices and AI-driven threats.
What is the need for Multi-Cloud?
This is exactly the kind of doomsday scenario that multi-cloud architectures aim to prevent. By distributing workloads, storage, and backups across multiple cloud platforms, you reduce the risk that a single provider's issue will take out all your critical resources simultaneously.
It's a strategy that aligns with the old IT adage of the 3-2-1 backup rule - maintain at least three copies of data, stored on two different media, with one copy offsite. Having providers in separate geographic regions further enhances resilience against localized disasters.
What is the difference between Multi-Cloud and Single-Cloud?
| Aspect | Single Cloud | Multi-Cloud |
|---|---|---|
| Vendor Lock-in | High risk of being locked into a single provider. | Reduced vendor lock-in, and the ability to leverage the strengths of multiple providers. |
| Resilience | Single point of failure, vulnerable to provider outages. | Increased resilience, and resources distributed across multiple providers. |
| Compliance | Restricted to the compliance offerings of a single provider. | More options to meet compliance needs across different providers. |
| Cost Optimization | Limited leverage for negotiating better pricing. | Ability to optimize costs by leveraging different provider pricing models. |
| Cost | Varies depending on the cloud provider. | Can be costly depending on the implementation. |
The importance of Multi-Cloud
So far, we've discussed how multi-cloud can eliminate a single point of failure in your deployments. However, this doesn't directly translate to how much you should care - and it's probably for good reason. Multi-cloud isn't going to solve all your scaling issues or cut your expenditure magically.
Perhaps the biggest benefit to multi-cloud, however, is preventing insidious vendor lock-in. In a recent report by Synergy Research Group, Amazon owns a staggering 33% of the cloud market share. This may not seem problematic at first glance, but an outage could impact a significant swath of companies.
The ongoing investigation by the UK's Competition and Markets Authority (CMA) into the cloud industry highlights several barriers to multi-cloud adoption, including technical challenges, high egress fees, and committed spend discounts that lock customers into long-term contracts. These barriers prevent businesses from freely moving between providers and taking full advantage of multi-cloud strategies.
According to Mark Boost, the CMA's intervention could lead to greater pricing transparency and competition in the cloud market, ultimately benefiting consumers by reducing costs and simplifying cloud service usage. Boost emphasizes that addressing these barriers is crucial for fostering a more competitive and resilient cloud ecosystem.
What are the benefits of using Multi-Cloud?
| Feature | Description |
|---|---|
| Risk Mitigation | Reduces the impact of a single cloud provider's failure or outage |
| Localization | Improves performance and user experience by hosting services closer to end-users in different geographic regions |
| Vendor Lock-in Prevention | Reduces dependency on a single cloud provider |
| Cost optimization | Allows organizations to select the most cost-effective services from different providers, optimizing overall expenses |
| Flexibility | Provides the ability to adapt quickly to changing business needs and requirements by leveraging the strengths of multiple cloud providers |
| Workload mobility | Supports the movement of workloads between different cloud environments based on performance, cost, or other business criteria |
Strategies for implementing Multi-Cloud
Executing a true multi-cloud strategy requires a lot of thoughtfulness and operational know-how. Simply having resources scattered across providers doesn't make you resilient. Teams need deep expertise in architecting and deploying applications for high availability across environments. Networking, security, monitoring, cost optimization - everything becomes vastly more complex at a multi-cloud scale.
But this isn't to say multi-cloud is impossible. Starting small by moving smaller pieces of your architecture, like storage, and replicating it across providers is a great way to get your feet wet. For example, companies could explore using a provider like Backblaze B2 Cloud Storage or Minio as an open-source alternative alongside their primary cloud storage solution, gaining redundancy and cost savings where it makes sense.
Moreover, egress fees - the charges for transferring data out of a provider's environment - could actively discourage companies from migrating off any one platform. With increasing volumes of data, these fees can disincentivize companies from moving data to another cloud environment. Mark Boost, goes into further detail about the corrosive impact of egress fees in a recent blog.
For Kubernetes users, a great way to start exploring multi-cloud is by testing the use of a service mesh for a multi-cluster approach, where each cluster is deployed in a different cloud provider. This allows you to gain experience managing distributed applications across clouds while leveraging familiar experience. Check out this piece on Multi-Cloud deployment using Crossplane for a deep dive into how you can provision resources across clouds using Crossplane.
Understanding costs and financial considerations
Understanding costs is another major hurdle when trying to implement a multi-cloud strategy. FinOps tools such as OpenCost and Komiser provide a unified platform for viewing your costs and projections of how much your spending will increase, helping you set realistic expectations and plan accordingly.
In an online meetup featuring Saiyam Pathak (Founder of Kubesimplify) and Jake Page (Community Developer at Tailwarden), they discussed how to take control of your multi-cloud environment by diving into advanced cloud tagging strategies. Watch the full recording here:
Summary
In this blog, we've explored the promises and realities of multi-cloud architectures. By distributing resources across multiple providers, organizations can enhance resilience and mitigate vendor lock-in risks.
If you are considering multi-cloud, be sure to critically evaluate your technical capabilities and whether enhanced resilience aligns with business goals. It's a commitment - but one that could pay massive dividends when the next big outage strikes.
Ultimately, whether the multi-cloud is worth the operational and monetary squeeze comes down to the maturity of your organization and applications. For companies going all-in on multi-cloud, it likely makes sense despite the complexities. However, for others, the costs may currently outweigh the benefits.
Further Resources
Looking to learn more about cloud costs and multi-cloud management? Check out some of these additional resources:
