Amadeus IT Group Speaks at the OpenShift Commons Gathering about Observability and OpenShift
Here in this blog, we are going to know what Amadeus IT Group Speaks at the OpenShift Commons Gathering about Observability and OpenShift.
The Past, the Present, and the Future: Towards the Cloud
On April 18, Red Hatters attended the OpenShift Commons conference in Amsterdam, which was held concurrently with KubeCon + CloudNativeCon Europe 2023. Red Hatters were joined by Augustin Husson (Principal Engineer) and Alban Hurtaud (Observability Architect) from Amadeus IT Group. The two discussed Amadeus’ journey toward becoming a cloud-native, open-source-driven company. Below is a link to the presentation’s complete recording.
With more than 690 million passengers expected to board aircraft in 2020 due to Amadeus and Navitaire technologies, and more than 646 million bookings managed and registered in 2019, Amadeus is the top technology provider in the travel sector. Given those figures, it should be no surprise that Observability-powered incident management is a must-have for Amadeus. Let’s examine the company’s migration to the cloud, which we will divide into four distinct stages as follows:
Amadeus began their journey to the cloud in 2016. At that time, the company relied extensively on logs and a self-developed observability solution, where metrics and traces were created by computing logs (for reference, see Figure #1). Engineers found it challenging to install the system from the ground up due to its complexity.
Given the complexity of the VM-based legacy system described above, Amadeus began making plans to move to the cloud. And that’s when OpenShift entered the picture. Four on-premises OpenShift clusters were spun up (using OpenShift 3.5) before Prometheus was installed on top of them. For the purpose of visualization, these four data sources were manually added to Grafana (see Figure #2 for reference). Despite these modifications, the time series continued to grow exponentially and required better management. Stage #3 began at that time.
In order to compare the response times of applications running in the cloud with those running on old VMs, the team also had to monitor Amadeus’ private data center and its historical workload. Amadeus was able to manage a 300 million time series in 2019 as a result of Thanos being added on top of Prometheus (see image #3 for a comparison).
Thanks to the adaptability and automated setups offered by OpenShift v.4.0, Amadeus will be able to set up 70 OpenShift clusters on Azure’s public cloud and 10 on-premises by the year 2021. Amadeus employs an operator to control the deployment of internal middleware, OpenShift clusters, its Azure subscription, and all of these clusters in order to deliver the apps. With the ability to run applications everywhere, both on public clouds and self-managed infrastructure (in this case, on-premises), OpenShift 4.0 truly became an enabler, enabling Amadeus to manage more than 1 billion time series events. To summarise, the Amadeus IT Group has adopted the motto “everything as code” and “automation everywhere” in an effort to create and scale its applications more effectively (see picture #4 for a reference).
Any Observability experience must include Working on a Single-Plane of Glass:
An Open Source Alternative Visualisation. Observability signals are typically visualized using Grafana as the industry standard. Although there are numerous backends currently available in the CNCF landscape, displaying Observability data is still an unfinished piece of the puzzle. In order to open source a single Observability visualization experience that can ingest a range of various data sources (traces, logs, and metrics), Amadeus and Chronosphere have been collaborating on Perses. Perses seeks to close this gap by supporting GitOps, enabling dashboards as code, and enabling users to embed charts and dashboards in any UI (for reference, see figures #5 and #6).
Amadeus is a significant player in the Observability market and has adopted open source extensively. Amadeus now has the option to grow and execute its applications everywhere it wants to by integrating with OpenShift, effectively controlling its workloads. Observability, however, has advantages for everyone. Because of this, Red Hat and Amadeus continue to work together on initiatives related to Observability after their work on OpenShift.