The Pros and Cons of Multi-Cloud Strategies for SaaS Applications

When I’ve gone about building new SaaS offerings, I’ve always been amazed by the passionate arguments by developers and dev-ops professionals about which cloud to select. It is the same with any technology selection, with everyone keen to champion their ‘home team’ and cement their value (or bolster their CV) by selecting the ‘right’ thing.

The increasing abstraction driven by infrastructure as code practices has created the belief that the future is ‘multi-cloud’, i.e. products will be cheaper to run and more resilient if we distribute deployment of application resources across multiple cloud providers.

So why would you want to do that?

The Pros of Multi-Cloud Strategies for SaaS Application Hosting:

1. Reduced Downtime and Failover: A multi-cloud strategy can help reduce downtime and failover, as companies can distribute their applications across multiple clouds. This way, if one provider experiences an outage, traffic can be routed to another provider to prevent downtime.
2. Flexibility and Innovation: Using multiple providers gives businesses greater flexibility regarding the features they can use. They can access unique offerings from each provider, which can help them innovate and differentiate themselves from competitors.
3. Reduced Vendor Lock-In: A multi-cloud strategy reduces the risk of vendor lock-in, making it easier to switch providers as needed, increasing flexibility and reducing risk.
4. Scalability and Cost-Effectiveness: Workloads can be distributed across multiple cloud providers, reducing the risk of performance issues and overpaying for excess capacity. The business is now in control of its application’s demands and can choose the most cost-effective option.

The Cons of Multi-Cloud Strategies for SaaS Application Hosting:

1. Increased Complexity and Management: A multi-cloud strategy increases the complexity of managing your infrastructure. With multiple providers to oversee, you’ll need to spend more time and effort integrating and monitoring them, which can lead to higher IT administration costs.
2. Higher Expenses: Using multiple cloud providers can lead to higher expenses, as costs will arise from supporting infrastructure and services. Additionally, migrating applications from one provider to another or developing custom integrations will incur cost.
3. Security Risks: Multi-cloud providers increase your attack surface. Data is more vulnerable when scattered across multiple cloud servers if you cannot easily define, deploy, and maintain common controls. In addition, security personnel working across numerous technology stacks will require additional training and support.
4. Sprawl: Multiple cloud providers can make maintaining control over deployed infrastructure difficult. An increasing burden of governance and process is likely to develop in response.

Is it a good idea or not?

Well, yea, but no. Of course, with anything, it depends.

Initially, I loved the leverage becoming cloud agnostic gave me in commercial negotiations with cloud providers.   The power to move workloads simply with ‘just a few code changes’ would see me optimising my cost to serve, driving up margins and lowering supplier risk.

Yet the reality of implementation would suggest that the cost savings achieved in infrastructure quickly diminished as the technical and operational overhead increased.

For example, we had built our security strategy around Microsoft Sentinel and on the assumption that ‘production workloads’ would be in Azure. A costing exercise to tool the SOC to monitor a multi-cloud deployment resulted in eye-opening costs. Eventually, the savings started to slip away as the increased skills required within the dev-ops and security functions to develop and manage the controls became evident.

Yes, multi-cloud would have delivered increased resilience and made our CVs look amazing, but with the modest spend at the time, it did not add up.

So is multi-cloud still a good idea, yes? Will it work for you? It depends on your goals. If resilience and avoiding vendor lock-in are your goals, crack on. If it is to save money… perhaps not.