Multi-cloud strategy: How to get started
Constructing a plan for working with multiple cloud providers? Let's ask some important multi-cloud questions
If you’ve been pondering a multi-cloud strategy but can’t seem to find “step one,” here’s a start: Let’s define exactly what “multi-cloud” means.
Cloud computing has generated more than its share of buzzwords and catchphrases – leaving many people confused about the difference between multi-cloud and hybrid cloud, for example.
Multi-cloud is a strategy, not a technology: It most accurately refers to an approach to deploying and managing cloud services. Multi-cloud really just means that you’re running two or more cloud services spread across two or more cloud providers. However, that doesn’t mean you have any coordination between the services or applications.
Hybrid cloud, on the other hand, “is a combination of one or more public and private clouds with at least a degree of workload portability, integration, orchestration, and unified management,” Red Hat technology evangelist director Eric Schabell notes.
“If your cloud model is missing portability, integration, orchestration, and management, then it's just a bunch of clouds, not a hybrid cloud,” he writes. (See the full article, 3 pitfalls of hybrid multi-cloud.)
[ Get lessons learned from your peers and cloud experts: See Hybrid Cloud: The IT leader’s guide. ]
More organizations are making the leap to a hybrid cloud model and a multi-cloud strategy - for several reasons, which obviously vary by organization. Your particular reasons matter – and defining them is actually “Step 2” in a nascent multi-cloud plan.
“Before you do anything, ensure all stakeholders are on the same page about why this is going to happen,” says Anders Wallgren, co-founder and CTO at Electric Cloud.
Good advice for any major initiative – though seasoned IT leaders know how often the answer to “why” remains elusive well into a project timeline. It’s particularly relevant here: Multi-cloud means multi-vendor, so from the get-go you’re adding potential complexity. On the other hand, you’re avoiding lock-in.
“Start with itemizing the reasons and benefits of leveraging a multi-cloud approach,” advises Amin Lalji, senior manager at EY Canada and a certified cloud security instructor for Learning Tree International.
Lalji points to a wide variety of reasons an organization might pursue a multi-cloud strategy: Vendor diversification, cost optimization, different proficiencies among cloud vendors, and data jurisdiction, for example.
Any of these on their own could justify (or even mandate) a multi-cloud approach; in many organizations, it might be some combination of business and technical reasons.
Wallgren shares his own brief, non-exhaustive list of possible answers to the question: “Why multi-cloud?”
- "We acquired a product that uses Cloud B and need time to port it over to Cloud A."
- "We have technical needs that aren't solved by a single vendor."
- "We need to make sure we're never locked into one cloud provider eventually."
- "We're moving everything from Provider A to Provider B and need time for overlap."
- "We can't use Cloud A in Geographic Location Z."
What are your multi-cloud use cases?
Once you’ve established your why, it’s time to flesh out the specifics of your use cases, in part because these become critical criteria in ongoing decision-making, from vendor selection to architectural design to staffing.
Avoiding vendor lock-in: It's a common multi-cloud scenario and, for some IT leaders, this might simply be a matter of practical habit (or of having been burned in the past.) Even here, however, you can more clearly define this goal as a part of a broader risk management strategy, for example.
For some CIOs, avoiding vendor lock-in might actually be a performance strategy, or part of a more sweeping plan to build reliability and resiliency into every system.
Let’s revisit Lalji’s other scenarios, too.
Cost optimization: Here’s a perfect example of a perfectly good initial motivation to consider a multi-cloud strategy that will ultimately need some additional thinking. Who doesn’t want to optimize costs? The hyper-competitive cloud landscape does offer opportunities here, especially in the IaaS and PaaS arenas.
“Major vendors are involved in aggressive price wars,” Lalji says. “Thus, the opportunity to move and shift workloads between vendors can significantly lower the total cost of operation for an organization that can actively manage their cloud footprint.”
But don’t let that delude you into thinking that running workloads in two or more cloud environments automatically begets cost savings. It needs to be highly strategic; if you’re playing “chase the lowest price” with every workload, you’re probably giving back cost optimization in the form of other resource drains.
“Cost optimization should not be approached as a ‘shift workloads back and forth between vendors’ strategy, as much as starting new ones with one provider and retiring workloads on the other,” Lalji advises.
Cloud vendor proficiencies: As with hybrid cloud, a multi-cloud strategy lets IT take advantage of particular capabilities or specialities of various cloud providers and environment types. This is essentially the payoff of the “not all clouds are equal” mantra. Yes, Lalji notes, there’s increasing convergence among what the major vendors can offer. But it’s certainly not uniform, so take the time to map your organization’s particular requirements to the proficiencies of particular vendors. Talking shop with your peers is one good way to get a stronger handle on those varying abilities.
“Some [cloud vendors] are emerging as front runners in specific services in terms of ease of use, resiliency of design, and overall performance,” Lalji says. “Organizations should choose to deploy specific types of workloads based on the vendor competencies while balancing requirements and costs.”
Data jurisdiction: For some companies, a multi-cloud approach may be driven – if not required – by the regulatory landscape, especially when it comes to public cloud providers. Such rules vary around the globe. Lalji shares one from his home base in Ontario, Canada, which requires that personal health data must be stored within the physical boundaries of the province. That rules out one of major cloud platforms, Lalji says, because the company’s Canadian data centers are located in Montreal, Quebec, a different province altogether.