Now that we’re entering the second decade of cloud computing, it’s time to revisit the prevailing wisdom about this IT deployment model.
In the early days—when there was essentially one public cloud—focusing your efforts on one cloud made sense. But now that there are so many capable cloud options, it’s time to set aside old metaphors and accept that multicloud adoption makes technical and economic sense.
Let’s examine facts versus fiction.
Myth 1: Nobody uses multimedia in production
incorrect. Almost every IT research firm says multi-cloud is a strategic necessity for businesses. In a survey, Gartner found that 81 percent of companies have a multicloud strategy, and IDC named 2021 the “Year of Multicloud.”
If you need specifics, look to companies like Mynet, which host popular online games. It uses seven clouds, including Oracle Cloud Infrastructure (OCI), to ensure high availability to its many addresses.
Or check out GoTo, the leader in remote work, which uses Kubernetes container management to run workloads on OCI and Amazon Web Services.
In the case of both GoTo and Mynet, the use of multiple clouds offers better flexibility and availability — if one provider is malfunctioning, operations can be routed to another company to keep customers online. This is the definition of intelligent IT deployment.
Myth 2: Multicloud costs more than a single cloud
No. Different public clouds offer a range of computing, networking, and storage services, and the prices of each can vary widely between providers. One of the advantages of increased cloud competition is that some service providers offer very competitive pricing for certain categories of service.
Remember: Just because one company claims to be the low-cost leader across the board doesn’t make it so.
If your business requires you to ship a lot of data to end users, then you should definitely pay attention to the “data out” or “outbound network” fees charged by different service providers. They accumulate exponentially (sometimes catastrophically) if demand spikes unexpectedly. One interactive video service saves 80 percent of data exit fees using OCI instead of AWS, for example.
Reducing the cost of data transmission between clouds will remove a major hurdle to multicloud adoption.
Myth 3: Multicloud is slow
Too false. While latency – or delays – can be an issue in data transmission, especially over long distances, this is less of a concern now that all cloud providers locate (or “share” in tech-speak) their physical infrastructure near each other. This proximity — in some cases different rooms within the same co-location facility — means millisecond latency between clouds. This, combined with the addition of more cloud data center regions around the world, means that the distances between cloud regions are shrinking.
In addition, some major cloud service providers have directly connected their own cloud. Oracle and Microsoft, for example, have collaborated on Oracle Azure Interconnect to provide fast flow of private data between OCI and Microsoft Azure. This enables many companies running databases and applications on OCI and other applications and analytics on Azure with a round trip latency of 1.2ms to 2.1ms. This is fast enough for most any organization’s requirements.
Myth 4: Multicloud is too complicated
Oh oh. This argument has been recycled throughout the history of information technology. Companies are told they must consolidate vendors to streamline operations and support. The problem is that different service providers excel in different areas, and over-reliance on one means that the customer loses leverage when it comes to purchasing power.
On the other hand, Oracle and Microsoft believe their customers are smart enough to handle “complexity,” as they have done for decades on premises, and are working to ensure that their common customers can use their applications of choice on the cloud of their choice. A key aspect of Oracle Azure Interconnect addresses this complexity by allowing companies to use OCI or Azure Active Directory for single access to resources running on either cloud.
Smart vendors can read the writing on the wall and will roll out more services to facilitate interoperability between the cloud. This helps many popular open source software packages – for example MySQL and Kubernetes – already run on all major clouds.
Multicloud management and monitoring tools will only improve and expand as more customers and providers adopt multicloud strategies.
The multicloud model will stick
The real truth is that as more company data and workloads flow into the cloud (I should say the cloud), business customers will use the services that best fit their needs, regardless of the provider running them. At the same time, companies will balance the price differences between the cloud on the underlying computing, storage and networking services. In particular, they will insist that fees for data flow between the cloud, or from the cloud to their customers, be affordable and free of fees charged.
Over the course of the relatively short history of cloud computing, we were told for the first time that no one would ever move anything to the cloud. After that, fingers were wagging that no one could transfer “production data” or strategic data to the cloud. Later, we were warned about the dangers of multimedia. It turns out that these oft-adopted certainties are fleeting because technology hates a vacuum and business leaders like having choices.
Today’s reality is different from the past, just as the future will be different from today. The only happy constant, it seems, is that customers will always have more and better options. Service providers who want to hold on to their customers should take note of this.
Leo Leung is Vice President of Product Management for Oracle Cloud Infrastructure.