CoreOS is all about Kubernetes and is the first (and only) non-Google team to date to head a release (Kubernetes 1.6). Kubernetes is unique in the sense that it’s probably the most difficult and most in-demand software in the enterprise right now, and that doesn’t happen too often. When a software has as steep a learning curve as Kubernetes, it’s often a very select group that shows interest, but that’s nothing like the current situation. The fact that enterprises large and small want to use Kubernetes to orchestrate its containers and no one knows how is where Tectonic steps into the picture.
Tectonic reduces deployment time in a big way by streamlining the management of applications through Kubernetes. Tectonic’s newest release —1.8 — is built around Kubernetes 1.8. The big highlight of Tectonic 1.7 was the general availability on the Microsoft Azure cloud, and this continues with 1.8. Before 1.7, Tectonic only fully supported container deployments on bare-metal servers and AWS.
CoreOS Tectonic: Supporting the hybrid cloud
A limitation of the new update, however, is that as of now users can’t run different operating systems within Tectonic 1.8 on Azure, so users need to either decide to deploy Linux or Windows container. With Microsoft and CoreOS both on the job that should just be a few wrinkles that will get ironed out before long. Additionally, Docker containers can run with a Windows operating system or Linux base and Microsoft has worked on providing support for Windows-based containers as well.
Kubernetes is what actually makes the hybrid cloud possible, which is one of its major emerging use-cases right now. Kubernetes does this by assuming the role of a master orchestration system and manages multiple datacenters, in multi-clouds and on-premises. While we all know the “multi-hybrid-cloud” is real, there are probably only a handful of people who can pull it off with Kubernetes. CoreOS is making hay while the sun shines and doing everything it can to ease customers into the Kubernetes learning curve.
Tectonic is always about building enterprise-grade features around Kubernetes capabilities and 1.8 is no different. It’s definitely worth mentioning that CoreOS is the first software vendor to provide users with painless automatic upgrades of Kubernetes. Being the first to come up with a one-click update for Kubernetes may not seem like a big deal until you’ve done a manual update or two on a live system. Additionally, the one-click upgrade comes without downtime!
Tectonic and Prometheus
To say monitoring has changed a lot since containers came into the picture would be an understatement. What used to be sipping coffee and waiting for a blip on the radar is now complete chaos if you don’t know what you’re looking for. Tectonic 1.8 continues 1.7’s integration with the almighty Prometheus, which seems to be the only tool up to the job of keeping a tab on containers at massive scale. Tectonic’s Rob Szumski was quoted as saying, "This has been one of the most requested features from customers.”
Tectonic had already integrated Prometheus in a collect-and-display mode in previous versions, but since Tectonic 1.7, the much-needed active alerting capability was added. Additionally, unique alerts for Deployments and DaemonSets were introduced to provide more visibility into the deployment process. With this ability, users are now alerted about their Kubernetes clusters via preconfigured alerts from Prometheus. Another feature since the 1.7 release is that customers can choose and configure the way they would like to be alerted.
Tectonic and Calico
You can’t expect a Tectonic release to not have some networking updates and a preview of a network policy feature to improve security and management were included. The Network Policy API moved to stable in Kubernetes 1.7, which is what the new feature is being built on. Network Policies make sure containers aren’t doing whatever they feel like and provide a strict set of guidelines with regards to communications between microservices.
These new network policies are now in alpha and powered by none other than Project Calico. Calico uses environment specific plug-ins to integrate with orchestration systems to provide networking between local and distributed workloads. This new update provides better security and control of inbound traffic to pods on Tectonic clusters. The new network policy makes sure you can control, block and audit network traffic.
There's a lot of mention of “upstream Kubernetes” with Tectonic, and what it means is that Tectonic pulls new Kubernetes updates extremely quickly. So by upstream they really mean the “freshest” version of Kubernetes, which isn’t surprising considering CoreOS and Kubernetes are practically family. The new release definitely targets developers looking to containerize their applications as sweetly and simply as possible while also targeting organizations looking to begin using cloud services.
Win-win for the cloud
Azure customers are definitely going to benefit on the time and money saved figuring out and messing with Kubernetes, not to mention the benefits of having your entire setup professionally built from the ground up. It all seems like a mad rush to get every customer to the cloud, and from AWS to Oracle to Salesforce and Google, that’s all that seems to be on anyone’s mind. Every vendor is trying to build the easiest and most user-friendly version of containers-in-the-cloud to entice users, and in the end, it’s the users who are benefitting the most.
CoreOS aims to make Tectonic the only Kubernetes platform to provide truly automated operations without getting locked in. Tectonic is now available across AWS, Azure and bare-metal environments and is free to use up to 10 nodes. The price per node after that is based on a lot of factors though the point here is you can try it for free to figure out if it’s your cup of tea or not.
A balancing act
Kubernetes-as-a-Service is definitely a big business, and if you can make it easier for someone else, more power to you. Kubernetes itself is getting easier to use day by day with improvements pouring in from all over the world (CoreOS included). Sounds counterproductive, especially since your product is based on the fact that Kubernetes is hard to grasp, but they’re obviously counting on big corporations that just couldn’t be bothered with any kind of learning curve at all.
It does look like CoreOS is in the driver's seat as far as enterprise-grade “upstream” Kubernetes goes and is doing everything to stay there. This industry, however, is changing every day and there are already a lot of options for orchestration. This includes smaller, lighter, and equally capable ones like Nomad. There’s still a long way to go before everyone gets to the cloud.