ArchitectureConference - Short25min
Cloud cost driven architecture at petabytes scale - BigLake vs BigQuery, which faster, which cheaper
This talk discusses the cloud-based BigData architecture that enabled the BigLake solution to outperform BigQuery, surprising even Google. It covers strategies for reducing storage costs, avoiding BigQuery slot reservations, and adapting to pricing changes. Lessons and best practices from Sabre's experience in optimizing data ingestion are shared.
Bartosz WieczorekSabre Poland
talkDetail.whenAndWhere
Thursday, June 12, 12:50-13:15
Room 3
In this quick technical talk I will be speaking about the BigData architecture in the cloud and how we were able to make BigLake solution faster than BigQuery - which was even a surprise for Google team.I will explain how we designed the architecture for storing petabytes of data in BigLake to reduce the storage cost in the BigQuery and how we achieved cost avoidance of buying BigQuery load slots reservation.Then I will focus on the pricing changes on Google side for BigQuery and BigLake and how it impacted our architecture around data ingestion. We had a couple of challenges after pricing changes to keep the cost rigor.At the end I will share the lessons learned and best practices from this optimization story based on Sabre production case.
Bartosz Wieczorek
For several years, in the Data and Analytics organization, Bartosz helped teams to create cost optimal architecture and fine-tuned several key Big Data Java-based jobs, resulting in substantial cost savings on cloud expenses. He collaborated closely with Google architects and engineers to identify cost-effective cloud solutions for Big Data processing. Bartosz actively shares his knowledge and experience at various tech conferences, including JDD, Google Cloud Data Live, GeeCon, and the Data Science Summit. Additionally, he has been a featured speaker at AGH University, internal Google tech talks, and Sabre internal and external meet-ups. You can find Bartosz regularly blogging about programming best practices on his personal blog and contributing to projects on GitHub.
comments.speakerNotEnabledComments