Getting ready for migration to Google Kubernetes Engine at Kiwi.com

The need

Kiwi.com - a well known Czech startup, operating groundbreaking search engine for flight tickets decided to migrate some of their systems to Google Kubernetes Engine (GKE). Sounds fun, right? But it's not easy at all. That’s why we were here to help.

The solution

We provided Kiwi.com with an on-site training for their Engineers. Our team prepared customised workshops focused on advanced GKE networking and scaling.

Kiwi's team was divided into small groups of twelve people. Each group had two lectors from Revolgy to provide individual attention to each of the participants.

The workshop was focused on:

  • Basics (such as application configurations and deployment, loadbalancing and probes, scaling, affinities, taints & tolerations),
  • Advanced networking (nginx-ingress and gce-ingress, communication in a pod, between pods, DNS, ingress & egress, firewall)
  • Advanced Workload, GCP and GKE (storageClass,PersistentVolumeClaims, StatefulSet, daemonSet) and
  • Usual Troubleshooting.

“The trainers were great. There was no question that would surprise them and they answered all of them in great detail and to the point”

says one one of the attendees

“Certain examples when the "hands on" approach became really interactive, example: setting up load balancers and liveness probes. The trainers were checking up, helping with issues and overall being interactive.”

another one adds.

To be able to concentrate more on the practical side of the workshop, our lectors gave attendees a small homework beforehand. They were supposed to go through a standard k8s basics course on qwiklabs.

“Thanks to good preparation of all attendees we could focus on more advanced topics such as http load balancing, persistency and horizontal auto-scaling etc.

Marek Bartík, one of the lectors and skilled Cloud architect from Revolgy

The result

After completing a series of workshops, Kiwi.com’s teams were ready to migrate to GKE. They are now ready to work efficiently with GKE at scale.