Making the case for hybrid/multi-cloud


In the beginning, convincing customers to adopt the cloud was a challenge. Only cash strapped startups would use the cloud because they couldn’t afford anything else. But then as the enterprise world started to see the financial benefits of adopting the cloud over running their own datacenters, the question became, which cloud vendor?

With the cloud wars fully heating up, with the big three (AWS, Microsoft Azure, and Google Cloud Platform) along with other major players (Oracle Cloud, IBM Cloud, Alibaba Cloud) and even tertiary vendors like Digital Ocean and Rackspace out there competing for business, there is still a huge untapped market still running on-prem or in leased datacenter space.

Unsurprisingly, most vendors want you to use all their products, keeping you entirely on their platform. That’s not new with the inception of the cloud wars. While working for 2 of the Big 3, I never understand why customers would want to run a multi-cloud or hybrid environment. Sure, there was an in-between period while you consolidated, but multi-cloud and hybrid environments can be overly complicated and data transfer fees between them add up.

I still think multi-cloud and hybrid environments CAN be overly complicated and data transfer fees between providers and locations are still a concern, but I have seen the light on why hybrid and multi-cloud solutions are not only here to stay, but for good reason.

Standards are great, that is why everyone has their own.

  1. Competition between vendors is good. It breeds better products and better pricing. And that’s a good reason to keep using multiple cloud vendors how, when, and where it makes sense.
  2. Local compute is still required for a lot of use cases, whether it be in a store to run local sales and inventory systems, or in a specific country because of data export laws.
  3. Each cloud vendor has tackled certain technical challenges in different ways, giving themselves advantages in some ways, and problems in others. Not every workload runs best or cheapest on your preferred platform.
  4. Not all cloud vendors offer the same level of infrastructure in the same places.

What I believe is the right way to think about approaching a hybrid or multi-cloud strategy is to break down your infrastructure and applications into clusters by who maintains them and how they interact with each other. Then break down who owns the applications from who maintains the infrastructure. Except for one strategy that I’ll touch on later, it doesn’t make sense for teams to become experts in multiple platforms. You didn’t let the Windows admins loose on your Unix systems and vice versa for a reason. The same goes for the cloud.

We also should be re-thinking how we write applications and deploy them. Containers, serverless compute and microservices are all the rage for good reason. When you break down your monolithic apps, you can start to house the different components where they make sense. That could be in a different cloud, on-prem running on VMware, or even different regions of the same cloud vendor. This is where you want to marry your DevOps teams with your platform experts, so the development and ops teams can figure out how to make the apps work, and the platform experts on how to make them work where you want them.

As you’ve broken down your datacenter infrastructure, you’ll see clusters of things that don’t really talk to each other. Those are the first candidates to evaluate where each would run best. Do POC’s and financial calculations to get a true comparison, but also don’t forget that cloud vendors are willing to give bigger financial incentives the more you combine on their platform.

You’ll also want to run these POC’s and financial calculations on the bigger clusters of applications too. You might find that it makes sense to run everything with one provider, or you may not.

You also need to look at what applications you run in-house or already with a SaaS provider you have and how they integrate with the rest of your infrastructure. G Suite and Office 365 are very popular and work well, but may not have the best native integration with the rest of your AWS stack for example.

You also need to invest in teams that understand how each platform works. Those core teams also need to know you’re investing in a hybrid or multi-cloud strategy and its not a competition between those teams, but the a healthy balance between the vendors you choose to do business with. They should be seen as one team with many arms or… more accurately, many SME’s. I’ve seen too many organizations spin up competing platform teams and things become hostile pretty quick. That’s not good for your business and could derail any benefit of your hybrid/mulit-cloud strategy if your people are infighting or simply not on the same page.

One way to avoid the former situation is to ensure you have a leader who understands each platform and its merits who can share the broader vision and rally everyone behind your business goals.

When most networking teams get involved, they are usually more concerned about network infrastructure and latency. You also need to make sure you’re taking network egress charges into account. Exporting large amounts of data from one location to another can be expensive, but sometimes it might worth it, so you really need to understand the broader pros and cons of this as well.

My last bit of advise… don’t treat a migration to the cloud as a datacenter move. Take the time to invest in re-inventing the way your IT infrastructure operates. Be disruptive and challenge the status quo. Don’t just “lift and shift” and think “oh, I’ll optimize later”. Use this as an opportunity to rethink how things get done and optimize for your own cloud journey.