The State of Kubernetes: Five Takeaways from Kubecon
This year’s Kubecon was the biggest yet on almost every vector: attendance, announcements and keynotes :). In between all of that were some stark themes on the state of the movement, the evolving vendor landscape and the challenges ahead. While there are some great summaries, we wanted to share our thoughts on what we saw and what we think the state of play will be in 2020.
- Kubernetes is more than a technology - it is a different approach to the build/package/deploy framework and is expressly designed for an environment of continuous change. It has rapidly cemented itself as the dominant computing paradigm due to its ability to abstract the physical infrastructure from the application stack in a way that facilitates the collaboration between development, operations, and IT. This is why entire companies (VMware, for example) are turning the ship to embrace the technology. Some commented the event felt a little “exuberant” but we disagree. There is real substance to support the hype.
- Kubernetes is designed for automation, but it is still not automatic. Kubernetes still requires considerable devops skills. That complexity is constraining adoption in some enterprise environments. This is an area where we expect to see tremendous progress in the coming year. It is an area we are actively working on from a storage perspective (stay tuned) and an area where we see considerable partner/vendor investment. Whether declarative or UI driven (or both), expect the next six months to deliver greater ease in configuring, deploying and managing Kubernetes - particularly around the proper security defaults. This will deliver another inflection point in its growth trajectory.
- Multi-cloud/Hybrid Cloud is the reality. While multi-cloud occurred organically, hybrid-cloud was always part of the strategy. Kubernetes is cloud-native. Object storage is cloud-native. Kubernetes wants to talk the S3 API whether on prem, on Google, on Azure, on AWS or any other public cloud (CSI only addresses block storage - not object and that is a larger discussion beyond the scope of this post - but we will write on it soon). While legacy storage companies are contorting themselves to look relevant in the cloud native world (mostly by using MinIO’s S3 gateway capabilities) the truth of the matter is that object is already the dominant storage class and will only get stronger. It is the only storage technology that can offer true, native interoperability with Kubernetes. Kubernetes has won. So has object storage.
- The edge continues to grow in its importance and is pulling Kubernetes out with it. There are a number of reasons for the growth in edge computing in a cloud native world, but the result is the same: storage, compute and networking must exist out there as well. This is a rich area for the vendor community and there is some very innovative work being done here - almost all of it taking advantage of MinIO. We even saw our logo in the Pure Storage booth where they were recommending us for the edge given our cloud-native footing and lightweight stack.
- The Kubernetes ecosystem continues to get more crowded - and with bigger and deeper-pocketed players. The IBM/RedHat merger was as much about Kubernetes as anything else. VMWare has bet the company on Kubernetes. Google, AWS, Azure are hyper-focused on making Kubernetes easier, more accessible, and more feature rich on their cloud vs. the competition. This will make life harder for the myriad of startups in the space. It is one thing to take out the legacy players, it is another to take out the establishment - especially the establishment which isn’t trying to “hold on” to outdated technology and business models. There will be winners to be sure, but there will also be losers and the speed will be different than other ecosystems. Remember, it was only a year ago when Docker pulled in $92M at a $1.3B valuation.
MinIO has benefited greatly from the Kubernetes revolution and that was on full display at the show. We estimate that 80% of our traffic came from current users. That is a stunning number that dwarfs even other “cloudy” places like Strata and Splunk .Conf.
It got us thinking and sent us into the data. Over the last month, nearly 20% of all MinIO configurations included Kubernetes. Over that month almost 40% of all MinIO instances were containerized. We are well on our way to having that be the default deployment configuration for MinIO. In the cloud native world, object rules and increasingly, MinIO rules object - particularly for the private cloud.
Odds are you already know how easy it is to try us out and to learn more. For the others, here is the roadmap. Download the code, check out the documentation and join the public Slack channel. If you want to talk to us, hit us up at email@example.com and we will structure a technical conversation.