The end of the cloud operating system?
Maarten Ectors
Innovative Technologist, Business Strategist and Senior Executive | Bridging Technology & Business for Lasting Impact
A bit over three years ago I worked for a company that produced the most popular cloud operating system. In 2020 I hope we do no longer need cloud operating systems.
?What is wrong with Cloud Operating Systems?
Technically nothing but if you do not have to worry about a cloud operating system, your life becomes so much easier. Google kicked off a “Kubernetes Revolution” in which everybody was invited to run software in mini containers in the same way Google runs their software. What nobody told the enterprises is that running a distributed Kubernetes cluster is hard work and needs lots of expensive experts. Even maintaining secure Docker images is very labour intensive and costly.
For the last two years we have been running serverless solutions in production and realistically you save a lot of money and headaches. No Dirty Cows to worry about. No scaling up and down to worry about. No army of expensive Kubernetes consultants. Just write some lines of code and let your favourite cloud provider(s) do all the heavy lifting. If something goes wrong then you have an SLA instead of nightmare. Several databases have also gone serverless.
Most enterprises will never ever have the scaling issues successful high tech giants have because most companies do not have customer problems and solutions that require zettabytes and nanoseconds. Technology companies are chasing a dream scenario which does not coincide with the reality of even top 100 companies.
So do enterprises really need cloud operating systems in 2020?
All about effortless control of your API ecosystem | Sync, Async, Multi-broker, Federation, ...
5 年A case of horses for courses perhaps? K8s managed in the way that some envisage should relieve a lot of the perceived admin burden and would allow firms to benefit from the additional flexibility they offer. In addition, regulated firms are increasingly concerned about vendor lock-in and serverless is perceived to make it hard to swap CSP(s).
Programmable Telecoms / Communications Expert
5 年Nice article Maarten Ectors. At TADSummit EMEA, https://blog.tadsummit.com/2019/11/22/tadsummit-emea-2019-summary/, serverless was one of the important issues raised by both you and Matt Millar?of?Updraft. It was also discussed in the open source telecom software track, its on the mind of many in the programmable communications industry. 2020 will see serverless grow in importance for those focused on failing fast, to discover successful recipes and then grow exponentially. Broader adoption will take much more time, this is enterprise IT we're talking about ;) I remember the phrase used by Matt of "Yak Shaving": a task, that leads you to perform another related task and so on, and so on — all distracting you from your original goal - of solving a customer's problem for which they will pay.
DevOps | SRE | Platform Engineering | Cloud | Linux | Blockchain
5 年Serverless is essentially an abstraction on top of Kubernetes. I think we will probably see cluster management increasingly move in to the hands of the cloud providers, much as we have with physical infrastructure.
CTO | AI & IoT Strategist | Building High-Performing Tech Teams & Scalable Systems
5 年I very much agree about designing systems for scenarios that probably will never materialise and the cost of such. I'm not convinced about serverless, maybe I'm old school.
Senior Staff, Site Reliability Engineering at Medallia
5 年Kubernetes isn't going anywhere for some time.