Private Cloud is Very Much Alive
I’ve read a lot of blog posts that either ask about or announce the death and failure of the private cloud.
While the public cloud offers unmatched elasticity and a slew of other benefits, the issue is no longer Public vs. Private.
The arguments for the fall of private cloud usually have to do with the problems of OpenStack. As OpenStack rises in popularity, more and more contributions are made, more projects are created and more companies try to pull the platform in different directions. The result is that in the case of OpenStack, popularity often equals complexity.
OpenStack has become a market all on its own, a hive of vendors and enterprises trying to figure out the best way to build a private, or hybrid cloud. In his blog post for TechRepublic, Matt Asay talks about the spectacular failure of private cloud and says:
“Private cloud lets enterprises pretend to be innovative, embracing pseudo-cloud computing even as they dress up antiquated IT in fancy nomenclature.”
He even goes as far as to reference OpenStack pioneer Randy Bias’ claim that Vanilla OpenStack doesn’t exist, as a testament that the platform is buried beneath distributions and complexity. I would say Randy’s point was missed.
While I feel it’s a bit harsh, I like the line quoted above. Not because it describes private cloud accurately — I don’t believe it does — but because it describes doing private cloud wrong.
The theme with the “private cloud is dead” arguments, is describing how OpenStack failed to provide enterprises with the agility and elasticity they need to innovate. While there are definitely testaments to private cloud failures, if you examine them a bit more closely, they’re usually testaments that doing private cloud wrong, will lead to failure.
I like making online videos, and have been doing so for quite some time. Every video I made during 2014 is either too dark to too bright. This doesn’t mean that my camera doesn’t work: it just means I was doing it wrong.
Let’s look at the results of a Gartner survey that was published earlier this year on Thomas Bittman’s blog:
140 attendees at the Gartner Datacenter Conference in Las Vegas where asked “What is going wrong with your private cloud?” Only 5% responded that everything is going great! Surely private cloud is the biggest failure since the latest Fantastic Four movie.
This chart and the blog post that came before it explain why some private clouds are failing. Using it to say that the concept of the private cloud has failed is misreading it. It’s true, failing to drive a cultural change, using the wrong technologies, having the wrong expectations, not having a defined strategy, are all things that can make your private cloud fail. We shouldn’t forget though, about organizations like NASA JPL, Yahoo! and Walmart, who embraced private clouds and made it work for them.
Based on the conversations we have on a daily basis with customers, prospects and the Scalr open source community, we see that private cloud remains an important piece of the cloud strategy of many enterprises. Now that many organizations had a chance to fail with private cloud, best practices emerge and are becoming more and more popular.
Here are a few characteristics of companies that leverage private cloud to their benefit:
Private Cloud is Just Part of Their Strategy
Earlier in this blog post I wrote that the issue is no longer Public vs. Private. The benefits of building a private cloud are different for each company and use-case. There’s a point to be made about data security and ownership, but I would argue that it depends more on your governance and security policies and how you apply them, than whether the cloud is private or public.
Whatever the benefit may be; HIPAA compliance, data protection, leveraging existing infrastructure, there’s an understanding that different tools are needed for different jobs. Trying to solve all of the company’s problems with private cloud will most likely not work. To borrow from Gartner’s Thomas Bitten once more, “Optimizing for everything means optimizing for nothing”.
A hybrid cloud solution, whether it’s a private cloud supplemented by a public cloud or the other way around, will give you the chance to tailor solutions to specific problems. For example, you can use your private cloud to tailor instance types to your workload to increase efficiency (something you might not be able to do on your public cloud). Or you can autoscale with your public cloud when expecting massive traffic spikes. Or you can create a complete solution with the right tools for notifications, budgeting, performance monitoring and so on.
One tool or one platform is not going to solve every possible problem.
A Cultural Change is Essential
If you’ll take another look at the chart we discussed earlier, you’ll see that 95% of the reasons for private cloud failure come from not having the right approach, understanding or expectations.
Building a private cloud is not always easy, but if it’s been properly assessed and found as the right solution for your company, the benefit can be great. Any company that adopts cloud, or any new technology needs to spend time and resources making sure all stakeholders come together to a partnership. I talked a little bit about this on Mission Critical: 5 Lessons from NASA JPL’s Cloud Journey.
They Use a Management Layer
Ok, there might be a bit of shameless self promotion here, but it’s true. One of the main benefits of cloud is getting things done quickly, and being able to easily track those things. Using multiple clouds is the way to go, but managing each cloud individually can be a waste of time and resources.
A unified UI and API, a templating and automation engine and global governance policies allow these companies to enable faster development, leverage each cloud for the things it does best and make sure IT retains control and stays within budget.
Private Cloud Doesn’t Have to Be Private
There are plenty of offerings out there that give you the benefits of private cloud without having to actually build one. Assess the problems you’re trying to solve by being clear on the reasons why you want to build a private cloud. Get a public cloud solutions architect involved, such as the excellent Miles Ward: there’s a good chance all of your needs can addressed by a public cloud. Or consult with @blueboxjesse because a hosted private cloud could do the trick?
Virtualization + Automation != Cloud
This can be attributed to the cultural change issue, but it’s important to understand that cloud is a service you offer to users. The existing Datacenter your company has is not a private cloud, and vCenter is still a few steps away from cloud.
That quote about dressing up antiquated IT in fancy nomenclature only applies if that’s what you’re doing. If a company decides to use virtualization and complex APIs to tackle developer productivity issues, their private cloud will indeed fail, on account of it not being a private cloud.
It’s understandable that the shift will take time, just like it did when virtualization became popular. We just need to make sure that once we’ve set our goals, we don’t attempt to solve our problems with the same tools that made us consider the move in the first place.
Making extravagant claims about the downfall of a technology or trend is good fun, and tends to get a lot of eyeballs as well. In this case however, I don’t think it’s called for just yet.
Over the last few years we certainly learned how not to do private cloud, which I’m sure will prove to be just as valuable of a lesson as the opposite. Cloud is one of the fastest growing industries around right now, and the private cloud is a huge part of that industry.
More and more companies are learning these lessons and building private clouds as stacked solutions. I’m looking forward to see how the private cloud evolves over the next few years, and it will be interesting to revisit this article then. I wanted to end this with a Star Wars reference about Cloud City, but I couldn’t think of one. Oh well.