KubeCon Retrospective: Kubernetes in Production

Antony Bursey | 30 November 2016

The Adoption of Kubernetes has been prolific since it’s v1.0.0 release in 2015. For many people it has been an intriguing technology for DevOps engineers to look at, making proof-of-concept builds. This year it became clear that Kubernetes is not just Production ready; the infrastructures have now been built and running deployments. Enterprises are now using Kubernetes to deliver services to customers and Start-ups are adopting it early to allow them to scale as they grow.

We had a look through all of the talks at Kubecon that demonstrate Kubernetes in production. It’s been great to see the way in which so many companies big and small are adapting Kubernetes to manage their infrastructure. It’s also a testament to the versatility of Kubernetes and the ingenuity of the engineers running it.

It’s a reality for many companies, both startup and enterprise that there will be large monolithic legacy applications. Moving to microservice infrastructures will always be inherently challenging, but we feel ultimately worth it. Dan Farrelly of Buffer gave the answer to their infrastructure problems- Kubernetes is the Tai Chi practising Patrick Swayze from Roadhouse who kicks ass and clears up the place. Their new microservice architecture built with Kubernetes provided stable application framework to allow their distributed team of developers to deploy code simply and quickly. Much like the swiftness of a roundhouse kick from Patrick Swayze.

Gif of an oiled up Patrick Swayze preparing to deliver a roundhouse kick

Startups often lack the resources for a proper DevOps team. Big enterprises on the other hand don’t. Over the past year we’ve seen numerous enterprises adopting Kubernetes to solve various obstacles to their architecture and make best use of their resources. Sam Ghods of Box explained they went from taking 6 months to deploy a new service to just under an hour with Kubernetes. That’s a 438000% improvement in deployment times for our Docker Swarm marketing friends.

Kubernetes in Chinese Enterprises - Xin Zhang, Caicloud

Full talk

“Kubernetes has rapidly evolved from theoretical trials to empirical deployments in an increasing number of US enterprises. However, the Chinese enterprises unveil different traits when it comes to requirements, platforms, and the tech-savviness of the operators, rendering the upstream guidelines and references a far cry from enabling successful Kubernetes production usage in varying circumstances.

Our unique experience with using Kubernetes to manage production systems in large-scale Chinese enterprises, with a stab at stereotyping different categories of common usage scenarios not covered by the official guidelines. Peering through the mist, we aim to glean insights into the usage patterns in different industries (carrier, finance, e-commerce, and traditional, etc) to use Kubernetes more effectively.”

For most enterprises, adoption of Kubernetes takes an a healthy investment in time and resources. But it’s clear companies are aware of the benefits. Justin Dean of Ticketmaster sees Kubernetes as the solution to their huge scalability requirements. Their infrastructure has to support demand spikes that go from idle to 100% capacity in only a few seconds. They currently have a team of six working on a goal to delivering a ticketmaster product into a production-grade Kubernetes cluster and equip their team with the skills required to support its operation. Rumors are each product delivery will incur a new internal operational convinence fee.

Ancestry is a similar story: new technologies like Kubernetes are a big shift for a number of traditional enterprise environments. They were your typical Microsoft .NET shop running in a traditional IT environment. It’s interesting to see a business with a heavy reliance on persistent data adopt containerization and microservices orchestrated by Kubernetes. Let us not forget in a world where this can happen, anything is possible.

Alena Prokharchyk, Rancher Labs, Inc. & Brian Scott, The Walt Disney Company

Full talk

“Many organizations run Kubernetes clusters in a single public cloud like GCE or AWS, and as a result have reasonably homogenous infrastructure needs. In these situations deploying Kubernetes clusters is relatively straightforward. Some organizations, however, have diverse infrastructure needs and as a result need to automate infrastructure deployment for Kubernetes across multiple clouds and data centers. This gives rise to a few challenges:

  1. How to ensure Kubernetes clusters in different clouds and data centers can communicate with each other, or in some cases even have a single Kubernetes cluster span multiple data centers.
  1. How to automate the deployment of multiple Kubernetes clusters.
  2. How to incorporate the new Kubernetes multi-zone clusters (f.k.a. Ubernetes) into multi cloud and multi datacenter deployments.
  3. How to manage the health of Kubernetes cluster itself, including, for example, how to detect and recover from etcd node problems.
  4. How to automate the upgrade of Kubernetes master and Kubelets, and how to handle multiple versions of Kubernetes clusters that exist in a single deployment.

In the last 6 months, we have worked with several enterprise IT organizations to solve these problems. We will share our experience on how to automate and simplify Kubernetes deployments.”

Fintech startup Monzo are in the process of building a banking platform with a micro services core powered by Kubernetes. Head of engineering Oliver Beattie described the performance and consistency requirements of a banking platform. Their infrastructure failover requires distribution across many physical data centres and multiple cloud providers. Obviously security and resilience are paramount for a burgeoning bank which sees the introduction of the latest CNCF hosted project, Fluentd to handle a number of service roles. It’s great to see Kubernetes bring some much needed stability to our banking sector.

Xoom: another fintech startup recently bought by PayPal have fully adopted Kubernetes. Xoom allows its customers to send money from the U.S to their families abroad and they are currently in the process of containerising all of their services in their production infrastructure. Also, we learnt the correct way of pronouncing kubectl. Please feel free to join us whilst we pretend this didn’t happen.

Next-Generation Microservices on Kubernetes - by Michael Kartashov, Bloomberg LP

Full talk

“Enterprise deployments are complicated. When managing proprietary technologies, sensitive client data and complex rules for access rights you inevitably arrive at a situation where your PROD environment diverges from your DEV and there is no certainty that your code will work in production. You wrote an update, your unit and integration tests pass, yet your cursor is still floating half-heartedly over the ‘RELEASE’ button. Sounds familiar?

DTP-on-kubernetes - the next-generation microservices platform at Bloomberg, allows you to run several versions of your microservice in parallel against the same requests, diff their output and trace messages through the system. On top of that, enjoy the deployment in seconds, brought down from hours.”

Kubecon Seattle has made one thing clear: the use cases for Kubernetes are far more diverse than they were 12 months ago. People are beginning to see its potential beyond simple deployment pipelines and development environments. As with all big organisational changes, they take time, but we’re starting to see Kubernetes in production; the proof-of concepts have been built and we’re now in the exciting stage of seeing how organisations are adopting it.

Need help running Kubernetes?

Get in touch and see how we can help you.

Contact Us