Is multi-cloud suitable for everyone?
To adopt a multi-cloud strategy and then use it to diversify/leverage the benefits of cloud infrastructure is not something new.
Currently, it's pretty standard to find out IT specialists that recommend a multi-cloud adoption, and consequently, companies that invest in this kind of solution. At some level, it's possible to say when it comes to a multi-cloud strategy that there's some hype around nowadays, highly commented and used. However, does adopt a multi-cloud strategy are the right thing to do and comes with the best cost-benefit to any company?
Let's put it straight: asking those questions doesn't come with the argument of saying that multi-cloud is a pointless fad. It's possible to think in multiples scenarios where this strategy is suitable for extreme performance and availability scenarios. I meant to shed some light on this article because multi-cloud adoption has its drawbacks - that are usually under-estimated in the decision-making frame.
This discussion's even more critical in the covid pandemic context. On the one hand, the pandemic has generated a global crisis. There is a reduction in the investment amount in new technologies (considering that recession has diminished the appetite for risks for general business and some industries). On the other hand, the pandemic - and business' constant adaptation to enable remote work - was a powerful fuel to adopt cloud infrastructure.
Those combinations of factors force decisions to be made in terms of investment in the cloud. This scenario should consider a multi-cloud architecture: its costs and, fundamentally, its fit the company needs.
SHIFT IN MULTICLOUD INDUSTRY
Before digging into this article's main topic, it's important to briefly comment on the industry's state of art current status. Also, how does this scenario impact the multi-cloud strategy meets reality. Initially, let's take a look at two magic quadrants below, produced by Gartner. They classify the primary cloud services available in the market as leaders, challengers, visionaries, and niche players. Take into consideration two indicators: ability to execute and completeness of vision.
The first graphic shows us a July 2017 picture. It's possible to notice a concentration in the niche quadrant—some competitors into visionary - like Google Cloud Platform. And AWS and Azure duo as incontestable market leaders.
Three years ahead of this frame, in August 2020, the scenario shown by Gartner's Magic Quadrant has changed significantly. Right away, it's possible to notice the market concentration has increased, with fewer competitors who had enough relevance to remain in the graph. While some services went away, just Tencent Cloud has emerged in Garnet assessment between 2017 and 2020.
There's one more significant change. The two quadrants - Visionaries and Challengers are now empty, and cloud services as Alibaba, IBM, and Oracle have now been classified as niche players. We can assume that these current requirements were significantly reduced due to an expected update observed in the market experience and also considering that the mindset of the market top talkers has also been changed. Given this scenario, I can reinforce the message that both issues were more crucial to determine this change than a loss of the value aggregated experienced by the business from a service perspective.
Lastly, this scenario has reinforced AWS and Azure, now with Google among them, as public cloud leaders. Even though Gartner still enhances the difference between them, the reality check is that they are becoming more similar. At least in a big picture perspective. There are differences in specific areas, but generally speaking, those three are very close in maturity level.
WHY IS MULTICLOUD STRATEGY NOT FOR EVERYONE?
The first reason to point out about multi-cloud strategy can as simple as a waste of money. If the leading players are way too similar, offering almost identical services - with similar prices. Some years ago, it was crystal clear the areas that each of them were particularly shinning out. For instance, Google was cheaper in storage, AWS always brought new tools/services as queues and cache, and Azure a substantial availability.
Those lines are not clear anymore. That doesn't mean that they are identical now. Quite the opposite, they still have differences. However, a company that choosing a multi-cloud strategy is naturally wasting the natural synergy that those platforms offer
Let's take queue, for instance. Each cloud has its specification but with its strengths and weakness. But, by adopting a multi-cloud strategy, the company will need a particular point of a generic approach to navigate the common area and get scale in gains. That means the company has two cloud services, but only using what they have in common, not their small special features that make a difference.
In this "pay 2, take 1" it's financially unsustainable, especially for startups and small companies. It can easily apply that same situation with queues in different levels of multi-cloud. So it's necessary to take the company's actual needs when it comes to multi-cloud.
One of the main points in favor of a multi-cloud adoption is to ensure the company's constant availability of services. So a multi-cloud environment, you should have virtually zero chance that two big players will go off simultaneously. But the tricky part here is: in fact, do you have an SLA agreement that you can't be offline X hours in a year? Can't you think about a strategy that avoids or mitigate a single point of failure in your architecture? Region availability in a cloud is easily configured. We're talking about a significant crash in AWS, Azure, or Google that all services became unavailable. It's easy to answer: "no, I can't be offline for a second. My services need to be 100% up.". But what are the structures on your cloud architecture that are a priority? In almost every case, you can work around the problem in the same cloud that doesn't involve so much energy, time, and money?
Any operational issue somehow affects the company financials and NPS. But there are few scenarios that a company can't work around on its architecture to avoid this headache. Maybe the question should be the focus on impact. Let's take a streaming video platform, for instance. To have two or three hours of downtime, it's a bad thing. But what if it's a banking service? Or a medical service? I bet that in the same cloud, you can see a near 99.99% SLA time.
And this impact is different from area to area, from industry to industry. However, these factors need to be considered when we think about multi-cloud, especially if we're talking about those big three market leaders.
WHEN MULTI-CLOUD IS THE RIGHT ALTERNATIVE
If a multi-cloud approach is not for everyone, some specific scenarios are viable or even necessary. The instances, however, are minimal. One of them it's about the moment that availability it's crucial, and cloud downtime can deeply compromise operational success.
Not only in a 24x7 medical scenario but let's imagine a ticket-selling platform that will sell tickets for Olympics or music festivals. With tickets being extremely high demand, it's a good practice to at least split the traffic between your clouds and keep selling - and commute fast - if one cloud goes down. But again, we're not talking about your application going down. We are talking about the whole cloud infrastructure in that region went away. Thinking about "what if it goes down, what should we do?" or "how can we redirect the traffic to another region or cloud?" it's the best we can do. We can assure that either everyone has the same access or no one buys - doing this, we can be seen by the public as fair as possible even when we had an issue with our infrastructure.
Another situation that it's recommendable to adopt multi-cloud is the security layer of banks or financial institutions. In those cases, companies were going for a private data center instead of a public cloud service. It was common to believe that it was hazardous running on a cloud when we talked about data privacy.
In those situations, multiple cloud structures might have a workaround. You can negotiate with the cloud provider a private cloud space that is not accessible as a public environment. Big financial institutions can now do this and allow their clients to be even more protected.
As I try to demonstrate in this article, to bet all in multi-cloud can be a valid strategy if some points are relevant among them company's financial capability to hire more than one service. But it's essential to get it straight: this approach can't be considered a standard or a market demand. It's possible to work with just one public cloud provider. Almost every time to go multi-cloud is to expend much money and not going deep into the best of the service can do.
That's why I state that it can't be considered a market trend to be followed by different companies in different industries. In reality, it is one of the many options that a company has available. And it considers things like cost-benefit, company needs, availability of SLAs, and data security.