My December LinkedIn blog “Private Cloud is Dead” sparked a lively debate. I got 9271 views, 584 likes, 116 comments, and 137 shares.
Many agreed with my core message: private cloud vendor solutions failed to deliver because they didn’t address the real pain points of customers.
What most customers actually wanted were SLA-backed private cloud services that met their performance, cost, privacy, security requirements with the same agility and ease of use as the public cloud but without the associated complexity or cost of building and operating their own private cloud.
Instead of solving this fundamental problem, many vendors sold their existing hardware – “cloud-enabled” tools, “cloud-washed” software, or open source distributions – and hoped that customers would successfully build and operate the cloud themselves.
But customers didn’t have the knowledge, skills, or people who knew how to build clouds, much less operate them.
Enter expensive and time-consuming professional services!
Six months and a million dollars later, you might have a semblance of a private cloud that is cobbled together by outside experts. But keeping this fragile and fledgling private cloud operating smoothly needed more budget and more skilled people!
The much-anticipated nirvana of increased agility and reduced TCO didn’t materialize.
Users defected to public cloud.
However, as many folks vehemently argued while commenting on my previous blog, the NEED FOR A PRIVATE CLOUD IS NOT DEAD. Far from it.
Enterprises are still looking for a solution. So, is there a better way?
To answer that question, let’s take a look at how the big three public cloud vendors – Amazon Web Service, Google Cloud Platform , and Microsoft Azure – tackle this problem.
Fundamentally, what they do is get software engineers to design and develop software to run cloud operations functions at scale. These Software Reliability Engineers (SREs)–as Google likes to call them–or more colorfully, “cloud Ninjas” take a software-first approach to automate everything that the operations and infrastructure team was doing manually. They are the ones who keep the massive public cloud engines humming smoothly in the background, and end users don’t even notice all the things that are going wrong and being fixed under the hood.
These SREs are constantly engineering “infrastructure-as-code” or “operations-as-code” or developing advanced distributed algorithms to automate every aspect of the operational spectrum: provisioning, installation, configuration, monitoring, application deployments, upgrades, rollbacks, emergency responses, failure handling, self-healing, auto scaling, proactive capacity planning etc. The list goes on.
Google even has a book on this. You can read it here.
Unfortunately, these SREs are extremely hard to come by. You need people with unique knowledge and skills in the areas of software development, large-scale distributed systems, infrastructure, and operations. The big three spend a lot of time hiring and training these types of people. You are not going to find them on the street looking for a job! So forget about building your own SRE team anytime soon. If you hire such folks for consulting, you are back to expensive cloud at a smaller scale that cannot compete with public clouds.
But there must still be a way around it. Right?
What if you had an SRE-like robot that could do similar things a human SRE does? What if it could learn about operational patterns, anticipate capacity needs, raise alerts about security anomalies, self-monitor and self-heal your private cloud in the face of failures, intelligently apply security patches and automatically upgrade hardware and software systems without any downtime? What if your private cloud could basically “drive itself” with minimal user intervention?
If such extremely complex machines as cars and trucks can self-drive themselves, why can’t private clouds drive themselves?
Sounds like daydreaming? Not really. These things are possible today.
Artificial intelligence (AI) to the rescue
Google’s DeepMind recently defeated legendary Go player Lee Se-dol in a historic victory just as IBM Watson wowed us all when it beat two of Jeopardy’s greatest champions. It’s not just about playing games though. Today, Watson analyzes massive amounts of visual data from medical imaging, like X-rays or MRIs, to predict diseases like cancer, diabetes, or congestive heart failure. Facebook instantly recognizes photos and suggests tags with nearly 98 percent accuracy. Personal assistants like Google Home and Amazon Alexa can carry out a natural human-like conversation and help us with routine activities like giving us audio updates on traffic/weather, online shopping, and even turning on/off devices at home! Salesforce Einstein is applying AI to improve sales and marketing productivity. The most game-changing technology we are already starting to see is self-driving cars.
All of this is real and happening today, driven by advances in artificial intelligence, natural language processing, deep learning, machine learning, large-scale distributed computing algorithms, and big data analytics. (See this HBR article by Aditya Singh of Foundation Capital: Deep learning will radically change the-ways we interact with technology).
If only we could apply some of these technologies to the data center, we could transform how companies build, operate, and manage their private clouds. It would give enterprises a viable private or hybrid cloud option that can compete and complement well with public clouds and help them lower their cloud costs, reduce complexity, and increase agility. This may be the only way to resurrect the private cloud from the dead and ensure its widespread adoption.
Wait a minute, such a self-driving private cloud is a fantasy, right?
Well, I wouldn’t be writing this article if this wasn’t real. Would I?
Such a self-driving cloud actually exists! It is being built by a Silicon Valley startup called ZeroStack. With their launch of an AI suite for self-driving private clouds, they are poised to disrupt the private cloud market and give the public clouds a run for their money. Read this blog by the ZeroStack CEO on how ZeroStack is building a self-driving cloud today: Self-driving cloud: From Vision to Reality
I’m also excited to share that I have joined the ZeroStack team to lead the product management vision and strategy. I believe strongly that an AI software strategy points the way to the future of IT operations. As customer workloads drive complexity in the datacenter, AI will be a key requirement for cost-effective management, and ZeroStack is at the forefront of using this technology.
Look to this space for more updates on how ZeroStack will continue to leverage AI to deliver a self-driving, fully integrated private cloud platform that offers the agility and simplicity of public cloud at a fraction of the cost. Having the private cloud self-drive itself through advances in AI and computer science allows enterprises to focus on their core business rather than cloud operations.