I have been prattling on about cloud topics for over six years now, helping IT practitioners come to terms with the changing world around them. It has often become a soul-searching, emotion-laden discussion.
Maybe I should get business cards that say "Cloud Therapist"?
In the last few months, it seems that cloud angst has started to reach an entirely new highs.
No such thing as a short meeting when someone needs to pour their heart out.
I think that's because -- when it comes to cloud -- most enterprise IT thinkers are waking up to the realization that they've hit an architectural wall, and it's starting to hurt.
It Didn't Start That Way
Go back many years, and I was explaining to people what a "cloud" was, and how it was different than traditional forms of IT architecture. Intellectually interesting, but hardly relevant to the world of enterprise IT back then. As an abstract concept, it was at a comfortably safe distance from day-to-day reality.
After AWS burst onto the scene, IT leaders were pressured to understand what cloud did, and potentially use it for some of their needs. Frequently, there was an emotional tone to the topic, as public clouds could be seen as competing with what in-house IT was trying to achieve on their own.
After that, a wave of interest in private clouds: flexible, cloud-like on-premises platforms that delivered infrastructure as a service (IaaS). All good.
To be honest, I saw most of the demand as a defensive reaction to demanding users who kept throwing public cloud services in IT's face, and demanding something similar.
And, of course, private clouds demanded a new operational model that created conflict with the rest of IT operations.
The seeds of discontent were sown when people started to attempt to have their private clouds work with public cloud services. Not surprisingly, the two were architecturally incompatible. And overcoming that incompatibility turned out to be horribly complex, difficult and expensive -- almost to the point of negating any benefits.
But it's 2015. And it's clear that cloud is the new architecture for enterprise IT. There's no turning back now, folks.
A Simple Picture
Imagine you've built a cluster of servers using your favorite hypervisor. Any application running on that cluster will essentially get the same operational experience. The cluster is managed and supported as a whole, and not piece parts.
We don't have a nice, flat cluster anymore, do we? We have two very different execution environments for applications. Two very different operational and management experiences. Two very different support experiences. And so on. Note: I'm not even getting into data movement issues here, just to keep things simple.
Assuming you can theoretically overcome the architectural incompatibility with infinite effort, we're not done yet. If a great degree of complexity is involved to move an application from public cloud to private cloud, or vice-versa, you're not going to do it too often. Plan with great care what goes where.
What ever happened to the vision of compatible hybrid clouds, seamless migrations, bursting and all that?
Vision Requires Execution
The enterprise hypervisor vendors (e.g. VMware / EMC / VCE) were able to deliver on a portion of the private cloud vision for IaaS, but never succeeded in delivering a meaningful compatible public cloud option.
IT shops who bet heavy are now at an architectural dead-end, it seems. Maybe someday, but when?
The public cloud vendors could deliver decent IaaS, but failed to deliver on a compatible on-premises option that works for enterprise IT. Bet heavy on a public cloud, and you're now stuck at the other end of the wire.
If we had the time and space, I'd share with you my thoughts as to why this situation exists, and why it's unlikely to change.
Which leaves enterprise IT leaders in a very, errr, interesting predicament.
The Frustration Can Boil Over
The business demands the speed and agility of public clouds -- especially when it comes to application development. And at the same time, the business also demands the control and economics of in-house private clouds.
It's long been true that the architectural decisions you make today will be the house you live in three years hence.
Three years ago, private clouds made sense, and no one was overly concerned about having a compatible public cloud option.
Or maybe it was OK to develop apps in the public cloud, despite the obvious difficulties of bringing them in-house.
Things have changed.
And the conversation can get pretty heated at times.
The Way Forward?
If perhaps we've made some architectural decisions in the past that didn't work out, how can we start making better decisions about the future?
First, acknowledge that the world has changed yet again. What might have made sense several years ago might not be the best decision going forward. Spill your guts, and move on -- the sooner the better.
Stop drawing little, incompatible cloud bubbles. Instead, draw one big compatible cloud landscape that incorporates private and public cloud working closely together. The better they work together, the easier your life will be -- and the more you will be able to do with cloud. And acknowledge that decisions made on one side of the wire impact the other side.
Third, begin to turn the ship around.
If you've invested in a technology that doesn't have a clear, functional equivalent on the other side of the wire, maybe it's time to migrate to ones that do. Yes, that process will take many years, but why wait?
No one is going to show up with a Universal Cloud Translator anytime soon.
The Oracle Perspective
You'll find a full, integrated IaaS/PaaS/SaaS stack, designed for enterprise IT. Many of the on-premises technologies Oracle sells (hardware, software, etc.) have compatible equivalents in the Oracle Cloud.
And, trust me, it's going to get even more interesting from here.
Maybe you didn't consider Oracle when building your private cloud. And maybe you didn't consider Oracle when selecting a public cloud vendor. Public and private clouds didn't need to work together, right?
But if you see your future as public and private clouds working closely together, using a consistent architecture and operational model, you'll find a uniquely interesting proposition.
And maybe I won't need to be a cloud therapist quite so often.
Like this post? Why not subscribe via email?