Multicloud - Understanding Benefits. Obstacles, and Best Approaches

Editor's Notes

• #6: Multicloud is one of the hottest buzzwords in enterprise IT today and is gaining ascendency as the approach to take when evaluating cloud adoption. This is supported by surveys such as the one taken at last year’s AWS re:Invent conference. <<CLICK>> No surprise that pretty much everyone surveyed is using the cloud and that almost all are currently using AWS. What may be surprising is that more than 50% are using Microsoft Azure and/or Google Cloud Platform. In fact, another recent survey indicate that the average enterprise self-identify as being users of 4 or more cloud providers.
• #7: But are we all clear on what we mean by the term, multicloud? There is evidence to show that some confusion exists in that area. <<CLICK>> For example, in RightScale’s most recent “State of the Cloud” report, 84% of enterprises state that they currently have a multicloud strategy. When digging into what that means, however, less than 20% of those customers indicate that are using multiple public clouds. Instead, nearly 60% of those saying they are multi-cloud have deployed a hybrid cloud, presumably with only one public cloud and one private cloud. This suggests that many users are conflating hybrid cloud with multicloud. But is that correct?
• #8: I am working under the assumption that it does matter for us to differentiate hybrid cloud from multicloud. Not because I want to be pedantic but because there is history and meaning behind these terms and conflating them makes it difficult to properly evaluate where we currently stand and how we should move forward. <<CLICK>> When we are talking about hybrid cloud, we have historically referred to an approach where a user is leveraging a public cloud with a on-premises/private cloud environment. This could involve running completely different applications in each environment or trying to run 2 instances of the same application in both environments. <<CLICK>> Multicloud, if we are to be precise, would involve the use of multiple public cloud providers to run a user’s workloads. This could involve running completely different applications in each cloud or trying to run multiple instances of the same application in 2 or more clouds.
• #9: I would argue that it makes more sense to talks about a hybrid multicloud approach where one or more private infrastructures are used along with two or more public clouds. As I speak with customers, I am seeing this becoming the norm and understanding that we are in hybrid multicloud world helps us to be more precise in how we identify benefits and tradeoffs for each architecture.
• #11: Typically, there are 2 overarching reason that companies take a hybrid multicloud approach.
• #12: Cloud bursting is one of those use cases vendors love to talk about but is rarely ever implemented beyond perhaps spinning up some web front-end servers across clouds. Anything beyond that becomes too complex and costly. If you have a 100 terabyte customer database that is used by your customer facing application, how do you make sure that it is accessible when you burst your application to another cloud? Do you replicate the data which is costly and could be time consuming? Do you keep the data on-premises How do you keep the replicated data consistent and up to date across all locations? How much would that cost? Network and security configurations differ across on-premises and across public clouds, how much effort and cost is required to keep that up to date for when you need to burst. Regional outages in public clouds do sometimes occur but multiple Regions or an entire cloud failing rarely happens. Given that there are way to architect for Regional failures, you have to ask yourself if it makes sense to spend the time and money to guard against a multi-region failure using a private cloud or a secondary public cloud? How many of you own a home and if you do, why do you choose to be locked in where the cost of getting rid that home is not cheap or easy? Users often talk about the fear that being locked into a vendor might mean that prices may one day be raised, which has not yet happened. This is where opportunity cost should be but is seldom is considered. Let’s say your costs double from one year to the next. But what if staying with one cloud means you can triple your revenues which lowering the cost of ownership, would that make a doubling of unit costs worthwhile? More on that later.
• #13: There are, of course, real reason whey a hybrid multicloud approach can make sense.
• #18: Use the right cloud for the right services to meet your requirements. That will often include services that would be too difficult and/or too costly to build yourself. One way to potentially work around the data gravity issue is to to store your data in a central location with high speed access to multiple public cloud providers. Won’t work for every data set and there are tradeoffs. However, this could be viable for data that you need to process using different applications and you want to advantage of different cloud providers and their strengths.