Is the end coming for cloud? Are we changing our view on centralized big-tech cloud computing in favour of decentralized, edge or bare metal?
Yes and No
Read this post to find out the five signs I think ‘the cloud’ is not the de facto target state for future workloads.
1. Slow Adoption
For many big enterprises, cloud adoption is a tricky beast. Governance and risk assessments, opening up to the ‘big bad internet’. Issues that maybe solvable in a technical way can be really tricky to solve in an organizational way. Think of processes that allow specific traffic in- or out of your network. Locking a platform like Microsoft Azure fully down, basically slows down the adoption of services and makes devops teams not fully utilize the cloud native capabilities of the platform, slowing adoption also, amongs the technical stack.
2. Vendor Lock-in
Well, we heard that before. But in fact there is a truth to locking in to a specific cloud vendor. For example when choosing a serverless technique, you are binding yourself to a specific vendor. I do not think it is bad per sé, but you should do the calculation of the costs of exit. Please do understand that any technique basically is a lock in on one level or the other. Choosing for Linux based containers for example, or for Kubernetes based on OpenShift is also potentially locking in to a vendor/technique. Think about what happened to CentOS just for a sec.
The now (in)famous Schrems II case makes legal departments run overtime to figure out how to deal with this case.
Customers of US cloud service providers must now themselves verify the data protection laws of the recipient country, document its risk assessment and confer with its customers.https://www.gdprsummary.com/schrems-ii/
Can we even use the public cloud for data of our customers? I’m following the outcomes of those discussions closely…
4. Edge Computing
The cloud (Google) basically brought us Kubernetes, but it also means you can run anywhere so also on the edge, on bare metal or in your companies private cloud. That means your ROI on existing hardware should be recalculated; you drive up density, run ‘new’ types of workloads and open up possibilities for your devops teams. I’m not saying running Kubernetes is easy but it does run virtually anywhere, the cloud is not the only place.
5. Risk, Governance and Compliancy
One of the most important blockers on using cloud seems to be this; how to manage risk, control governance and ensure you are compliant. Being and proving you are ‘In Control’. And yes, most cloud providers have well-defined policy frameworks and control toolsets, but in practice this is a pretty big subject that requires time to setup properly in an entire enterprise. Think about banks, investment companies and insurers; they cannot just ‘do what they want’ because the legal implications of doing it wrong are enormous. The slow down caused by this issue can be a burden for devops teams simply giving up and ‘eating their dogfood’ i.e. in the form of a private cloud or DC.
Is it all doom and gloom?
I’m a avid fan and proponent of cloud computing. I truly think Microsoft Azure, Amazon AWS and other cloud vendors can help you gain speed, be more agile and even save costs doing it well.
Every now and then a post like this surfaces, so I’m guessing in 5 years time, we will still have ‘the cloud’ thankfully. But it is always interesting to reflect on choices and watching for changes in the landscape of computing technologies. Also; try to think about cloud in other ways than only the technical possibilities. Think about the ramifications for compliancy and risk.
Any ideas or feedback? I’m open for a (virtual) cup of coffee anytime!