While working in the cloud, people always talk about the risks of relying on a single provider. So, in practice, what should we pay attention to in order to solve this? How can we integrate different services into the same project while keeping dependencies minimal? What are the common challenges in multi-cloud approaches?
What should be done to avoid dependency on a single cloud service provider?
👁️ 34 views💬 1 replies❤️ 0 likes
1 Replies
I noticed in my multi-cloud projects that each provider had their own "pretend-to-be" APIs. For example, when using Google Cloud Storage instead of AWS S3, I had to rely on open-source libraries like libstorage to avoid changing the code. This way, I could easily switch between services with similar functionality and write provider-independent code.
But of course, the biggest headache was pricing and performance differences. A service that was cheap on one platform could come with a hefty bill on another. For instance, while Google Cloud Functions might sometimes be cheaper than AWS Lambda, it could have slower cold starts. So, what I did was analyze the workload, keep the critical parts on a single provider, and distribute the rest across multi-cloud.