Aditya Jain
SDE from Expedia
Actions
A small introduction about myself, I have been working in the DevOps fraternity for 7-8 years & have been part of the Expedia group for almost three years gaining a lot of experience. I am fascinated by how the DevOps community is growing day by day, I have had the privilege to attend DevOps Days conferences in the past years, and continuing with the same mindset I am looking forward to various opportunities to share my journey where we resolved bottleneck issues for my organizations as well as gain knowledge around how international DevOps community is solving their respective problems.
Navigating Traffic: RCA on connectivity timeouts, Service Controller Migration and Its Impact on EG
This case study focuses on the RCA done by Expedia to figure out the reasoning behind delayed requests, connectivity timeouts, then controlling traffic flow, configuring istio for enhanced performance, & migrating the service controller.
Key points
In this session, embark on a comprehensive exploration with us as we:
Delve into how we pinpoint the underlying root cause to be the abrupt termination of worker nodes, during node churn controlled by istio service controller.
Figuring optimal solution catering to our platform involving k8s & aws community collaboration and deciding on the network load balancer IP type service controller utilizing pods as targets instead of nodes.
Mitigate risks by rigorously testing a vast array of scenarios, metricizing uninterrupted traffic flow, and connectivity performance improvement.
Expertly overcome challenges to seamlessly migrate from istio service controller to AWS LB controller in a high-scale production environment ensuring zero downtime.
Leveraging Keda for event driven customised scaling of applications
DevOps transformation up to today's world has made a lot of transitions and one of the major situations to be resolved today is applications scaling, not just the infrastructure which has its own solutions but the underlying application scaling based on custom metrics. Expedia has employed Keda and how it resolved a great number of cases.
Expedia being very rich in terms of use cases, has various scenarios where the standard metrics scaling was not fruitful to scale applications as intended. We solved this issue by employing Keda scalers in such a manner that the developers would be able to leverage their custom metrics to scale their applications without needing much knowledge about Keda functionality.
We would love to share this journey with the whole community about shortlisting the right toolset and how it solved a number of our problems and made the dynamic scaling of applications more efficient, robust, and justified and guide them about walking on the same path which could be in line with our roadmap or something individual organizations would like to devise on their own from our story.
How Expedia Group built Infrastructure as a Service architecting DevOps solutions with AWS Karpenter
We would like to share our journey across the DevOps community regarding how expedia solved a major infrastructure concern in today's DevOps world which is dynamic autoscaling leveraging open source toolset karpenter. We started using Karpenter in its very early stages and embarked upon a great journey involving a lot of architectural solutions customized to expedia's use case and a lot of feedback to the AWS karpenter community regarding improvements.
Contribution to the community: We faced some very good challenges as karpenter was a very promising but an immature tool when we picked it. Leveraging an opensource tool backed by AWS in such a huge platform like Expedia had a lot of challenges, learnings, and decision making. We would like to share this with the whole community about our rich experience and guide them about walking on the same path which could be in line with our roadmap or something individual organizations would like to devise on their own from our story.
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top