Three reasons why multi-cloud is back - and here to stay

Profile picture for user Dominic Wellington By Dominic Wellington January 26, 2020
Summary:
MongoDB’s Dominic Wellington explains why multi-cloud has come back into fashion. However, this time there are new rules that apply and as a strategy for IT decision makers, it’s not going anywhere...

I’ve worked in IT long enough to spot cycles. And like me, ideas will come and go.

It’s like synthpop. Everyone fell out of love with it when the 80s were done, and now it’s back, stronger than ever and spawning new genres like synthwave.

Personally, I’m all for that particular comeback, but the tricky thing is, nobody knows how long each cycle will last.

Parking 80s music for a moment and looking at our own industry, one of the biggest IT strategies today is multi-cloud.

This too was a cycle - it fell out of love and now it’s back. But this one is different. It will be no cycle. It’s going to stick around, and in this post I will outline the three reasons why:

     1. Someone else is doing it for you

a. Outsourcing complexity is the reason people want cloud in the first place. Multi-cloud used to bring the complexity of integrating different clouds right back to end users, which is why it never took off in a big way. This time, the complexity is abstracted away – which is how it should be. Happy days.

     2. Use all of the features you need

a. Locked-in no more. Multi-cloud lets us benefit from a variety of providers to reap the rewards of varied capabilities across multiple jurisdictions. Think local pricing and differentiated features. 

     3. Run anywhere (really)

a. Do you want to spin up an application in India while you are NYC-based? No problem. At least one of the big clouds is going to be there, and that means that your service can be too.

Image of Dominic Wellington MongoDB

It’s different now

The previous multi-cloud debate took place in a world in which a “server”, whether physical or virtual, was still the main building block of IT services.

That's not so relevant today.

These days, while there are undeniably still a lot of servers, layers upon layers of abstraction separate services from the infrastructure they run on.

In this context, multi-cloud starts to mean something different.

When implementing a new business service means working with serverless components and the APIs that communicate with them, the fact that those components might span multiple clouds flips back into being an asset, because end-users of the service don’t have to deal with the integration themselves. The vendor of the business service is negotiating the differences between the various cloud services and only exposing the parts people actually care about.

And before you know it, synthwave is a thing, and multi-cloud comes into fashion again.

Back in fashion.

1. Someone else is doing it for you

First of all, one huge benefit of serverless to users is that they do not need to manage the complexity themselves.

If a provider in the stack elects to run their service across different cloud providers, this is (or should be!) transparent to end-users.

The benefit is that experts are working out how to deliver the best performance for your workload, across all of these different options, without you having to work it out for yourself.

2. Use all of the features you need

Secondly, this modern version of multi-cloud flips the idea of being locked-in to one service.

Instead of being limited to specific features of one cloud in one region, working on a multi-cloud basis allows a business to reap all the rewards - wherever they might be.

Take working on a preferred platform with a SaaS component like MongoDB Atlas - our cloud database service. We offer MongoDB Atlas across AWS, GCP, and Azure, so businesses can set up projects on different providers to take advantage of specific pricing schemes, local regions, or platform features.

Without multi-cloud, this wouldn’t be possible. If you are using one provider for your core infrastructure, but a different one has a cool machine learning offering that you want to use, without multi-cloud capabilities that would be a big issue, if not a deal-breaker. If you need to work with a partner that made a different infrastructure choice in the past, one or both of you would have a lot of painful work ahead of them – unless you had opted for a platform that offers multi-cloud support, and deals with the complexity for you.

3. Run anywhere (really)

Finally, far from locking users into a specific cloud provider, an abstracted service can follow users as they migrate between providers, or mix and match providers within their operations as circumstances demand.

No need for rushed migrations after an acquisition, or for duplicated services and expenses.

Need to launch in a new country or on a new continent? No worries – at least one of the big clouds is going to be there, and that means that your service can be there too.

So multi-cloud is here to stay

The user is back in control.

They can pick whichever cloud provider suits their requirements in terms of functionality, geography, or policy compliance, and even move between them in a single environment.

In this context, the historical objections to multi-cloud notably around expense and lock-in are as relevant as discussions of CPU architecture.

Today, we all care much more about rolling out our new game, website, or shopping service than the intricacies hidden at the bottom of the stack.

After all, isn’t that what the cloud is supposed to be all about?

Multi-cloud is back just like synthpop is and it's time we all got on board.