Multi-cloud is increasingly becoming the norm. You might imagine that this has something to do with the benefits of multi-cloud: the ability to avoid lock-in, the power to mix and match, the apparently competitive pricing, or the lack of a single point of failure.
It’s a reasonable assumption to make, but it’s actually not the case.
Most businesses don’t make a purposeful, deliberate decision to go multi-cloud. They aren’t persuaded by the benefits. They end up using multiple clouds, more often than not, unintentionally.
In most cases, a business unit, frequently as a result of slow service delivery from IT, has purchased services from a cloud vendor without the involvement of central IT. Fast forward a few years, and core IT has caught up and implemented a different vendor resulting in a multi-cloud deployment. Over time, the customer retrospectively justifies this position as “strategy” rather than “accident!”
Whether intentional or not, businesses should also be aware that a multi-cloud strategy – for all but the biggest consumers – is often a false economy and introduces more complexity and cost than benefit.
1. Duplication and fragmentation
One notable disadvantage of the multi-cloud approach is duplication and fragmentation of data, services and skills. This results in a higher cost of operation for storage and data transfer. Multi-cloud environments make it hard to track which applications are running, where your data is, and how much this all costs you day to day – which in itself, leads to compliance headaches. When you do decide to shift data around, there are data egress charges to consider.
Managing a multi-cloud environment often results in compromises and increased cost. Typically, the native tools within the cloud vendors meet most requirements, but once a company chooses to move into a multi-cloud model they need to purchase third-party tooling that often isn’t as feature rich as the native tools and costs more. Alternatively, organisations may persist in using multiple cloud-native management tools, which lead to fragmentation and inconsistency in policies and a lack of a single source of truth for compliance reporting.
Moreover, each platform will have its own set of variables that make optimisation a real challenge - billing systems, pricing models, instance/VM sizing differences, data egress fees, and so on, that will make budgeting and cost control more complicated.
Even if storage and app sprawl issues are somehow ironed out, multi-cloud businesses are still going to need to make multiple investments in network connectivity, firewalls, and so forth – so duplication is almost an inevitability. It’s also highly likely, regardless of how well you manage the environments, that your cloud bills will rise, as volume discounts are slashed because usage is split across multiple providers.
2. Human capital
Another problem to consider is the inefficiencies created in the people who manage all of this. The diversified nature of multi-cloud means that staff who would usually only be expected to know one public cloud have to get to grips with many.
Unfortunately, this usually means one of two things: either expertise is spread thin and your business has to spend additional time and resources to upskill staff in these different areas, or the teams have to split in order to focus on one particular cloud and have minimal crossover. Neither is particularly desirable.
In short, the more clouds there are to manage, the more in-house technology skills must be developed and maintained. The problem is - most companies don't have the skills to manage infrastructure across clouds in general, let alone container applications that are strung out across cloud providers.
3. Not making full use of public cloud solutions
If you want to go multi-cloud, interoperability is key – businesses should be avoiding cases that require reconfiguration of apps, as well as any feature disparity. Unfortunately, to achieve this, you have to resort to the lowest common denominator between platforms.
Standardisation is constraining. Organisations often end up running virtual machines and containerised apps, and not really unlocking the true power of public cloud which are the in-built native services.
Given those three problems, is multi-cloud ever a good strategy?
The answer entirely depends on what you’re trying to achieve. What are your aspirations and why are you going to the cloud? If it’s to increase agility, global reach, and access new services then you can do that perfectly fine with one or more.
Generally multi-cloud works best in larger organisations where there are autonomous units that can design, test, build, release and support their own solutions and services. This allows those units to work with the vendors and tools that best suit them and allows them to maintain their specialisation.
A less common approach is to look at this by technology platform, e.g IaaS in one location, data platform on another location. Again, the downside is data egress moving between the vendors.
If you find your business spread across multiple cloud services, don’t panic. This doesn’t make things unmanageable, but nevertheless, you need to formulate a clear strategy to ensure you’re unlocking the value of multi-cloud.
While it’s certainly possible to let individual business units or teams decide which public cloud to use, make sure you don’t end up kicking the can down the road and making the infrastructure more complex, duplicated, and difficult to correct at a later date.
The more workable approach is to create a preference for one public cloud over the other. Across the whole of the business, one should become a primary and default cloud service, while the other should serve as a secondary cloud system.
This secondary cloud must only be used when there is a clear business case. Perhaps there are custom apps that have been built on the cloud technologies, and there is little to no ROI in recoding them to the new platform.
Which public cloud you choose to preference obviously depends business-to-business, and on each business’ affinity. Existing relationships, in-house skillets, and fundamentally, how each provider is engaging with you and supporting you on that journey will be the determining factors.
So, to summarise - in the words of the American economist and business strategist, Michael Porter, “the essence of strategy is choosing what not to do”. It’s an apt rule for IT departments to live by.
Yes, there are circumstances where multi-cloud can work well. But generally speaking, a successful cloud computing strategy is about making difficult choices and trade-offs, and selecting certain platforms over others. There most certainly is such a thing as having your fingers in too many public cloud pies.
Sean Roberts, General Manager of Public Cloud, Ensono
Image source: Shutterstock/bluebay