Ogury and the challenge of meeting growing analysis needs while optimizing costs

To meet the challenge of analyzing more data without exploding infrastructure costs, Ogury used Spot EC2 instances which enabled it to reduce the costs of big data EMR clusters by 57%.

start a python tutorial
kr_quotemark
From the start of Ogury, we quickly experienced the limitations of our old cloud provider and we experienced some scale issues. Our product is deployed on mobile phones around the whole world, and if they are unable to contact our servers, we cannot analyze and study the data to transmit our advertising campaigns. Which is of course a problem for our development "
Olivier Bazoud
Head of infrastructure of Ogury

Ogury and the challenge of meeting growing analysis needs while optimizing costs

The team wanted to scale their infrastructure by keeping up with the very rapid development of the business while controlling costs and freeing themselves from basic infrastructure management.

When scaling and cost optimization go hand-in-hand

Ogury analyzes tens of terabytes every day, from 400 million profiles worldwide who have consented to share this data. These are filtered, normalized, and analyzed in a datalake. The lessons learned from these analyzes allow for advertising activation. Faced with the need to optimize the cost of running these actions, while scaling, Ogury turned to AWS. By the summer of 2015, Ogury had completely migrated to AWS.

With AWS, everything is already ready. For example, we very quickly used Amazon EMR . We were starting to analyze a large volume of data, and we needed this service to avoid managing big data clusters ourselves and to save time, ”explains Olivier Bazoud.
Switching to AWS, Ogury wanted to do a lot more in speed and volume, without increasing expenses proportionately. A lot of work on optimizing the services used allowed them to achieve this goal.

Analyze more data, without exploding infrastructure costs

Amazon EMR allows Ogury to analyze data, with a first step of data cleaning: cleaning the data, removing duplicates, normalizing the data, aggregating them. Secondly, with machine learning or statistics, Ogury deduces higher-level information, such as age groups, which will help with the targeting of campaigns.

All data, several peta, are stored in Amazon S3 in managed service, which gives Ogury a reliable service, and a certain ease of use since Amazon EMR reads directly from this storage space. The products are installed on Amazon EC2 instances, in microservices or back-office. "The use of Amazon EC2 Spot Instances has enabled us to achieve significant savings, reducing the costs of our big data EMR clusters by 57%! », Notes Olivier Bazoud.

Ogury collects and analyzes the customer journey data of their users by combining with noSQL Amazon DynamoDB databases for consent. “We use it as a managed database, which saves us time,” explains Olivier Bazoud. We can focus on helping developers, without having to optimize the database. It is paired with the traffic, which makes it possible to scale to hold the load ”.

Amazon Redshift is Ogury's datawarehouse, and makes it possible to query this data, carry out studies or know the targeting of campaigns for certain users (who are anonymized).

Finally, Ogury archives its data on Amazon S3 Glacier and uses the Amazon EC2 Reserved Instances to, again, reduce costs. “We have also worked a lot with AWS to optimize costs on Amazon Cloudfront . The data that we analyze uses a lot of bandwidth to reach us, and this has a significant cost in the process ”.

Responsiveness and improvement of time to market

“AWS supported us in our scale. It would have been complicated to do without them and even more complicated to do it ourselves on premise for example. We would not have had the same speed of implementation. However, the development of the company went very quickly, we had to keep up with the infrastructure. There was only AWS to keep up, ”said Olivier Bazoud.

Ogury is also improving its time to market. About a hundred employees work in R&D, with a product whose heart is based on technology. “Using AWS allows us to go faster than our competition. This is continuous improvement, we are in agile mode. The development team releases new features every two weeks or more. " For Olivier Bazoud, Ogury needs the responsiveness of AWS services.

Maintain growth objectives thanks to a solid infrastructure
“To manage this infrastructure across 4 regions, with huge traffic, the team is quite small. We support the growth and scale of Ogury, and this is what AWS allows us to do with confidence. There hasn't been a time when infrastructure has slowed the growth of the business. It is no longer an obstacle to the development of a company as it may be the case with old models, or before there was the cloud and the deployment was counted in months. If Ogury has the scale and the growth we see today, it's because the infrastructure has held up to this exponential growth. "


About Ogury

Ogury is the world leader in mobile branding advertising. Created in 2014, this Adtech startup is now present in 10 countries and employs more than 400 people. She works with more than 1,500 brands and 3,500 publishers to build a trustworthy digital advertising ecosystem based on user choice. Ogury Advertising Engine is an integrated set of technologies, from consent management to engagement, designed and optimized for mobile branding campaigns.

Summary

Amazon EMR

Amazon EMR is a leading cloud-based big data platform dedicated to processing large amounts of data using open-source tools such as Apache Spark, Apache Hive, Apache HBase, Apache Flink, Apache Hudi, and Presto.

Find out more

Amazon S3

Amazon Simple Storage Service (Amazon S3) is an object storage service offering industry-leading scalability, data availability, security, and performance.

Find out more

Amazon EC2

Amazon Elastic Compute Cloud (Amazon EC2) is a web service that provides secure and scalable compute capacity in the cloud. Aimed at developers, it is designed to facilitate access to cloud computing resources at web scale.

Find out more

Amazon DynamoDB

Amazon DynamoDB is a key-value and document database, delivering latency performance on the order of milliseconds at any scale.

Find out more