The enterprise is quickly transitioning from a single-cloud environment to one in which workloads are balanced over multiple clouds. But while this represents a dramatic shift in enterprise infrastructure, and is certainly not without its management challenges, many organizations are finding that the benefits far outweigh the concerns. What is needed is a clearer understanding of what multi-cloud architectures entail and how they can best be leveraged for emerging workloads.
Here, then, are the top 10 myths surrounding multi-clouds:
Myth 1: Multi-Cloud Data Management Is Complicated
The fact is, multi-cloud architectures can be managed through a single interface, which makes them easier to orchestrate than today’s silo-laden legacy infrastructure. As Avere Systems’ Scott Jeschonek notes, many enterprises are using network-attached storage (NAS) to accelerate the integration of legacy systems to object storage platforms in the cloud. In this way, compute resources can access data directly from any source, perform their operations and then send data back to storage either in the data center or the cloud.
Myth #2: Multi-Cloud Equals Hybrid Cloud
Red Hat’s Radhesh Balakrishnan summed it up nicely to The Enterprise Project when he pointed out that a multi-cloud consists of clouds hosted by various providers, while a hybrid cloud is a mixture of public and private resources. This is an important distinction because it can greatly influence the ways in which the enterprise can allocate workloads across the appropriate resources. In general, all hybrid clouds are multi-cloud, but not all multi-clouds are hybrids. (To learn more about these different types of clouds, see Public, Private and Hybrid Clouds: What's the Difference?)
Myth #3: Multi-Cloud Is Less Secure Than Single Cloud or On-Prem
While it is true that, as Barracuda Networks says, multi-cloud pushes the concept of shared responsibility for security to new levels, most security platforms are already incorporating this element in their latest releases. New licensing options are making it easier for the enterprise to meet both application- and data-layer security requirements, while dedicated links running through a secure cloud aggregation portal allows network security protocols to be enforced across the full cloud ecosystem.
Myth #4: Multi-Cloud Is More Secure Than Single or On-Prem
This isn’t to say that multi-clouds provide an added layer of security, of course. According to IOD Cloud Technologies Research, most multi-cloud architectures to date suffer from high degrees of fragmentation across infrastructure, tooling and cultures. The more disjointed the pieces, the more attack vectors are present, forcing the enterprise to implement increasingly complex security regimes to lock them down. As mentioned above, however, an integrated orchestration stack goes a long way toward mitigating this problem.
Myth #5: Multi-Cloud Management Is Best With Open Source
It seems logical that an open management platform like OpenStack and CloudStack will work with a larger pool of cloud providers than a proprietary one, but this must be weighed against the added internal resources and skillsets that come with open source. And in many cases, proprietary solutions support the APIs of the leading open solutions anyway.
Myth #6: Multiple Clouds Are More Expensive
On a per-GB basis, multiple clouds can cut costs because the enterprise has greater leeway to shift loads to the most efficient architecture. Companies like Rackspace are offering a range of multi-cloud service tools that encourage users to shift workloads to the providers of their choice, even to rivals/partners like Amazon and Microsoft.
Myth #7: Multi-Cloud Is Only for Large, Established Enterprises
Small businesses have specialized applications, too, and a single provider is not likely to deliver optimal support for all services. ImageKit.io’s Somesh Khatkar notes that many start-ups benefit from building applications around a multi-cloud strategy in order to alleviate migration and integration hassles later. And since many providers offer free service tiers for low-scale operations, it’s a good way to start generating revenues without incurring significant upfront costs. (To learn more about migration, see What Moving an Idea to the Cloud Actually Entails.)
Myth #8: Enterprises Embrace Multi-Cloud Only When They Are Ready
Few organizations have been able to avoid shadow IT, so chances are your data may already be on multiple clouds without your knowledge. Meta SaaS CEO Arlo Gilbert says this is dangerous because without a clear understanding of where and how data is being stored, the enterprise runs the risk of exposing data to theft or losing track of it altogether, which will affect the quality of analytics and other functions. And it puts a kink into cost-control measures as well.
Myth #9: Multi-Cloud Is Optional
Technically true, but only in the sense that building a successful business model is optional as well. As WhirlWind Technologies’ Maliha Balala notes, digital transformation will likely require diverse and widely distributed infrastructure, and no single cloud provider – not even Amazon – can deliver optimal service for all data and applications. Multi-cloud also provides an idea sandbox to test proof-of-concept experiments.
Myth #10: Multi-Cloud Is Worry-Free
Even though multi-cloud breaks the cloud lock-in paradigm, it usually requires locking in to a single management platform. Also, as tech consultant David Linthicum points out, not all cloud APIs offer full-service compatibility, leaving the enterprise with only subsets of common features from each provider. In-house management and brokerage tools might also fail to update on a regular basis to accommodate new services at the speed at which they are introduced.
It is important to remember that simply moving data to multiple clouds is not the same thing as having an integrated multi-cloud architecture, and devising a seamless, optimized environment is harder still. As the enterprise gravitates toward a more diverse cloud ecosystem, maintaining data and application portability across all clouds should be a core consideration. Otherwise, you run the risk of building the same silo-based infrastructure over the wide area that currently hampers performance in the data center.