It is the same story in any industry: startups are leveraging innovative technologies to outcompete long-term incumbents. These “digital attackers” deliver a new level of customer value by looking at an often long-existing problem from a different angle. Interestingly, we find digital attackers in every industry, from traditional (e.g., consumer banking, insurance, retail) to newer verticals such as enterprise software, biotech, and space exploration.
Netflix was one of these digital attackers. Today, the company must make sure to remain in “attack mode” since otherwise a younger and hungrier attacker will eat their lunch.
Digital Attackers Eat Incumbents’ Lunch Even in Commodity Markets
Banks are not able to compete by offering higher interest rates on savings, cheaper mortgages, or other significant monetary differentiation. However, cashing checks via smartphone app is an offering that makes everyone’s life a lot easier. Who likes to go to the bank to cash or deposit their expense check, paycheck, bonus check, or any other kind of check? Mobile deposit saves me about 30-40 minutes for each roundtrip to my bank, so I personally save approximately 3 hours per month that I can now spend playing with the kids. Another example is a mobile app for barbershops. Personally, I gravitate to the two major chains that show me a nice Google map with little clocks on it that indicate the wait times for the different locations. There are other barbers that do a better job at cutting my hair. However, the difference isn’t significant enough to justify a long, frustrating, and unproductive wait, which always adds to the LTE bill.
Today, Every Company is a Software Company
It all starts by understanding that every corporation, old or new, without exception, must transform into a digital attacker. “Sorry, but we have only just enough resources to keep the lights on, so let’s talk about this again tomorrow,” is no longer a viable answer to the CEO’s request (or demand) to build new software to better serve customers. So what are the CIO and CTO to do? The answer is simple: “Shift resources away from keeping the lights on and toward proactively developing and validating new services.”
Shifting From 50/50 toward 90/10 – No More Herding Pets
Even modern data center environments that offer developer self-service and server configuration management based on declarative management tools are still stuck with provisioning servers as pets instead of treating them like cattle. This means that even when Chef, Puppet, and/or Ansible are used, there are always different servers and server groups that “need” different configurations and have different operations management requirements. This approach is what left us with the 50/50 resource split between keeping the lights on and focusing on innovation.
OpenStack was Supposed to be the Solution
Five or so years ago, OpenStack was hyped up to be the solution for all of these problems, and basically create a NoOps cloud just by plugging USB sticks into commodity pizza boxes. However, it turned out that you cannot just gloss over the lack of infrastructure automation and standardization. OpenStack projects turned out to require a lot more implementation homework than anticipated. And since existing data center staff had not done this work in the past, they were not able to do it for OpenStack either, so enterprises were left with the choice of hiring high-dollar specialists or giving up on private cloud for now. Many chose the latter and went to AWS and Azure instead.
You could compare placing a private cloud, even an elegant one, on top of data center infrastructure that is not well automated to putting lipstick on a pig (photo credit to: BeDifferentorbedead.com)
OK, Then Let’s Move Everything to AWS or Azure
EMA research has shown that public cloud cost is the key infrastructure-related pain point for enterprises to resolve in 2018. AWS and Azure resources are easy to consume, but they are difficult to govern, especially when there are dozens of separate accounts in one enterprise. Developers often make the convenient and safe choice in terms of instance type and size, and they often are prolific in terms of how many resources they create but forgetful when it comes to turning off these resources or downsizing them when their project is complete. Even the very popular EC2 Container Service (ECS) is based on preprovisioned EC2 instances that customers pay for whether or not they are used.
So What’s the Answer? We Need a Private Cloud 2.0
We need a private cloud solution that leverages commodity data center infrastructure, as well as Azure, AWS, Google, etc. resources to automatically serve up DevOps APIs and standard DevOps tools, based on cattle, not pets. This means that whenever anyone in the organization needs a development environment, additional test instances, or demo instances for the biz dev guys, this private cloud will carve out the required resources based on a cost/risk calculation and send users the access details for the new environment.
In my next post, I will explore how machine learning and AI are critical components for this type of hybrid cloud.