Lessons Learned on Cloud Tech/Costs
By Steve Libbey
For the last 15+ months I’ve had 30 or more conversations with senior executives in the enterprise IT space. These enterprises include smaller and mid-size companies up to household name Fortune 500 corporations. Our conversations were about cloud technology and economics. Surprisingly enterprises of all sizes had shared interests and concerns.
Here are the lessons learned on cloud tech/costs:
Cloud native interest coupled with fear. Literally everyone is talking about the cloud and either how to get there or how to stay there more efficiently and cost effectively. Enterprise IT organizations of all sizes are interested, fascinated even, by cloud native plans, processes, technologies, and platforms. But they are equally terrified to start a cloud native journey. Most organizations (70%) have not yet started the cloud native journey but of those that did the overwhelming majority (79%) failed. There have been quiet conversations about now ex-employees who tried and failed to go from monolithic to microservice applications and go cloud native. How to mitigate the operational, financial, and technical risk of a cloud initiative is the overriding concern by enterprise IT executives. They want help getting to the cloud safely, securely, and economically but with flexibility.
Vendor lock in. Horror stories abound about binding cloud financial contractual commitments or inflexible cloud tools that only work in one environment or vendor. Names withheld to protect the guilty. I’ve heard repeatedly from enterprise IT execs that just training teams on new tools and platforms is too time consuming and financially onerous. The execs that I’ve spoken to sometimes refer to the cloud providers as being like the Hotel California, where you can check in but you can never leave. Many of the tools and technologies around the cloud are focused only on one specific cloud provider. Enterprises all said that they want commonality across tools and processes and more standardization of solutions and portability between cloud providers.
Predictable costs. If there is one hot topic to avoid (that isn’t politics) its cloud spend. If you’re mean spirited and want to see an IT exec foam at the mouth just ask questions about cloud spend trends. The cloud toll is costing CIO’s their jobs and getting the CFO’s out of their offices and into the CIO office for painful and pointed questions. There needs to be a better way to more accurately gauge, predict, and leverage cloud costs for the overall betterment of the organization. The alternative is increasingly bitter recriminations between enterprises and vendors and efforts to repatriate. I know of two companies just this week that are leaving the cloud. Completely. Out of frustration with a 7-figure annual cloud bill. There needs to be a smarter way to use the cloud more cost effectively. Cloud competition would be a great place to start and an overall rationalization of cloud tools and platforms.
Conclusions/recommendations based on my conversations:
Use a DevOps platform that is cloud and Kubernetes neutral. There are platform solutions out there that can be effectively used with any cloud provider or flavor of Kubernetes. A DevOps platform is a time and cost saving way to quickly start a cloud native project or program. Why do it all in-house? I know of a Fortune 200 senior IT executive who started his own custom Kubernetes project. He had to hire 10 kubernetes certified technologists who were very difficult to find and cost him a lot (~7M over 3 years). The project still isn’t complete. Leverage an existing platform to mitigate time and financial risk. Decades ago most companies out sourced the payroll department to ADP. Then travel went to Concur, HR went to Workday, and CRM went to SalesForce. Does it really make sense to build your own customized Kubernetes engine when there’s alternatives out there? Make sure that your platform is cloud and K8s agnostic.
Leverage open source. There are over 300 CNCF recognized open-source projects out there with Kubernetes being the first, biggest, and most well-known. These open-source projects have varying levels of maturity and market acceptance. Use the most common and most useful open-source Kubernetes applications for your cloud native project(s). There are three reasons for this; 1) You can benefit from thousands of engineer hours of development time. 2) The bigger open-source applications have more attention, usage, and adherents. 3) By using open-source you always have the flexibility of bringing everything back in-house.
Fixed consumptive models. There is widespread revulsion from enterprise IT executives about any commercial offering that starts free and ends up costing you your house. I’m not petitioning here for capped cloud pricing but rather advocating for a tiered pricing model with associated levels of technical support and capability. There is a massive market opportunity, based on what I’ve heard, on fixed price cloud solutions that include pre-determined levels of consumption and correlated levels of included technical support and functionality. Think of it as Gold, Silver, Bronze. No more open-ended pricing models is the market demand. The market is headed in this direction and will get there in some way in my opinion.