Is multicloud actually worth it? ☁️☁️☁️ (gathering perspectives for an article)
James Hirst
1y
Most large enterprises I've seen are multi-cloud, at least in part due to acquisitions and legacy infra: They buy a business and want to integrate, but it will take time. The acquisition are on GCP, the core business is mostly on AWS, but also still has legacy stuff on their own infra/private cloud and it's going to be 5 years until they move off it. And let's not talk about the RevOps team in a different building who've built everything on salesforce and Azure... 😀 Everyone has a consolidation strategy, with clean lines, single supplier, efficiencies of scale. Then they meet reality. Life's messy, business is messy, embrace it and have a strategy to handle it.
Worth it or not it is a reality. When you ask for worth, is it around monetary facts, integration, needs...? True cases that you find: 1. When acquisitions happen and one company was an AWS shop and other was Azure shop. 2. When you have a SaaS offering and the customer wants to host the data in their cloud. I have seen control plane in AWS and deployments in Azure, AWS, GCP and Alicloud. 3. The fact that you need to be ready for those scenarios on top, as a software company, makes you needing to have that multicloud setup in place. A more technical scenario - having an API Gateway to reverse proxy AWS lambda functions https endpoints but the identity provider being Azure and not AWS Incognito. Hope this helps
There are some exceptions potentially but overall being "single" vs "multi" seems like a recipe for disaster in terms of your need for (future-proof) flexibility, scalability (up or down), cost & time, engineering efficiency, integration... Providing "fit for purpose" private/cloud/public/hybrid/SaaS options to fit customers' business & technical needs is most needed in uncertain 2023 times
Bill Doerrfeld here's a great article on how Snap Inc. thinks about multi-cloud: https://www.protocol.com/enterprise/snap-microservices-aws-google-cloud
Axel Grosse
1y
my 2 cents Did you ever disagree with a vendor about his next years pricing? If yes then you know why a multi-vendor strategy is paying off on the long run. That is especially true when you talk about your infrastructure and applications in the cloud.
I see a lot of chatter and two particular real implementations of multicloud. One is a sort of composite architecture, bringing components from multiple providers to one place, either to get the best tools from all the clouds, or because they somehow ended up with a weird combination (usually through acquisition as the other commenters have mentioned). The other is organisations choosing cloud-agnostic services deliberately, so they always have the choice of moving. Few do, but it feels like a really smart business move to be able to - I'm not sure if I see a disproportionate number of this second type of implementation due to working mostly in open source space where the risks of lock-in are better understood.