Delivering Added Value For The Ad Trading Desk, With A More Efficient Infrastructure

How to lower costs and improve scalability with Google Kubernetes Engine

By switching to Google Cloud, Gamned is reducing its infrastructure costs by 30% while improving productivity, so it can focus on developing the best programmatic ad tools for its clients' campaigns.

GOOGLE CLOUD RESULTS
  • Enables the IT team to focus on developing value-added tools, by automating and simplifying infrastructure maintenance
  • Supports the rapid deployment of new features with simpler development and deployment processes
  • Helps improve targeted campaign generation with rapid analysis of data daily using BigQuery

MORE TO INVEST IN INNOVATION, WITH 30% LOWER COSTS

What’s your favorite browsing device? A smartphone, tablet, or even your smart TV? To adapt to the many different ways consumers now engage with content online, digital advertising has become more sophisticated. It’s not just about ads reaching consumers, it’s about how and when they reach them. Brands need to tailor ads to suit the platform they’re being delivered on, as well as personalizing the content they’re delivering.

As an early pioneer of programmatic advertising, which uses software to buy advertising space, Gamned has witnessed this evolution over the past decade. Its research shows that it’s crucial to reach target audiences at the right time in their digital journey, and to present a message that’s adapted to their expectations in terms of tone and values. To help brands do that, Gamned enables its clients to create programmatic ad campaigns using banners tailored to specific social media platforms, as well as audio and video options.

In programmatic advertising, we need to be able to react quickly in order to win new clients and to build successful, dynamic campaigns. We felt we'd hit a wall with our previous infrastructure, and wanted to spend more time on creating value. That's what migrating to Google Cloud offered us.

— Jonathan Marcopoulos, Technical Director, Gamned

Gamned’s teams work closely with its clients to meet their goals, whether that be targeting existing customers or finding new audiences. The most common model is for Gamned to run the campaign on the client’s behalf using platforms such as Google Display & Video 360, although it also offers training and support if clients want to develop their own skills internally. « What sets Gamned apart is the fact we work across so many platforms, » explains Jonathan Marcopoulos, Technical Director at Gamned. « We bring together the data from across those campaigns to offer clients a unified snapshot of how their ads are performing. »

To power campaigns for the more than 200 brands it works with, Gamned processes large volumes of data every day. It has multiple internal platforms and portals to help its internal teams optimize campaigns to target audiences accurately and deliver as much added value as possible. Previously, this infrastructure had been spread across multiple providers, with some services hosted on-premises. « Gamned had grown through various takeovers without a clear infrastructure strategy, » explains Jonathan. « We needed to centralize control, with the goal of better managing our costs and removing barriers to faster development. »

After comparing three potential cloud solutions, Gamned chose to work with Google Cloud Partner OP-Rate and Google Cloud, as it was excited about the possibilities offered by switching to a Kubernetes-based architecture on Google Cloud.

« In programmatic advertising, we need to be able to react quickly in order to win new clients and to build successful, dynamic campaigns, » says Jonathan. « We felt we’d hit a wall with our previous infrastructure and wanted to spend more time on creating value. That’s what migrating to Google Cloud offered us. »

MIGRATING TO MICROSERVICES MEANS GREATER PRODUCTIVITY

Working with multiple providers can multiply the operational workload. With a number of different hosted services, Gamned’s IT team was spending too much time managing providers and keeping track of their different pricing structures. So that its team could spend more time on developing internal tools, it decided to find a single platform to host its services with a focus on gaining better control over costs by exploring on-demand pricing.

During a comparative tender with three implementation partners in summer 2019, OP-Rate’s suggestion, which involved switching to a Kubernetes-based model, caught Gamned’s attention. « We were already using BigQuery and so we’d had a taste of what the cloud could offer us, » explains Jonathan. « OP-Rate showed us the benefits of switching to microservices, speeding up our deployment process by using tools such as Terraform, and creating a CI/CD pipeline. »

« OP-Rate’s approach is based on collaboration, » says Marjorie. « We see ourselves as an extension of the IT team we’re working with, and worked closely with Gamned to define the goals for the project from the start. » Working with OP-Rate, Gamned was also able to benefit from account credits to cover its initial migration period, which meant it avoided doubling up on costs and didn’t have to rush the migration.

In February 2020, once OP-Rate delivered the new infrastructure, Gamned began by migrating two internal web portal services to the cloud. The idea was to test out the new tools and give its developers time to learn more about Google Cloud. After six months, the team felt confident in developing the infrastructure further and undertaking more complex migrations.

Now we use Google Kubernetes Engine for all Gamned's workloads including web portals, data processing, and recurring tasks, so it's a lot easier to manage our environments. That gives our developers confidence and makes them more productive, as they can deploy new features much quicker

— Jonathan Marcopoulos, Technical Director, Gamned

DEPLOYING NEW TOOLS FASTER WITH A SIMPLIFIED PRODUCTION PIPELINE

Gamned’s services now run on Kubernetes clusters and Compute Engine VMs, orchestrated using Google Kubernetes Engine and using Pub/Sub to automate data ingestion. For its storage needs, Gamned uses Filestore for high-performance low latency workloads, and Cloud Storage for standard and coldline storage, with Cloud SQL as its relational database.

Previously, Gamned’s development, quality control, and production environments all had very different deployment processes. That made it difficult to make sure they mirrored one another, which was inconvenient for the development team.

« Now we use Google Kubernetes Engine for all Gamned’s workloads including web portals, data processing, and recurring tasks, so it’s a lot easier to manage our environments, » says Jonathan. « That gives our developers confidence and makes them more productive, as they can deploy new features much quicker. »

CRUNCHING EVEN MORE DATA, EVERY DAY

Before the migration, Gamned’s data services were spread across multiple data storage and analysis tools, which meant its teams were spending time on converting data between formats and money on data transfer costs. It was already using BigQuery and producing reports and visualizations with Looker Studio, but now data from all the platforms Gamned works with is centralized in BigQuery.

« I’ve tested other similar tools, but none of them achieved the same performance as BigQuery, » adds Jonathan. « We use it for our large data analysis tasks, and it saves us a lot of time. And the developers feel really comfortable working with SQL, a language they’re all familiar with. »

As it was recently acquired by the TF1 Group, Gamned will now have access to larger quantities of data from across the group’s online editorial portfolio. Its new data infrastructure will help it to manage these larger volumes of data as it grows.

Migrating to Google Cloud has given us the flexibility and control we needed, as we're able to choose on-demand billing where that works best for us. Since migrating, we've reduced our costs by 50%

— Jonathan Marcopoulos, Technical Director, Gamned

FOCUSING ON INNOVATIVE TECHNOLOGIES TO DELIVER CLIENT SATISFACTION

According to the initial goals set out with OP-Rate, Gamned has successfully met its cost-reduction targets, as well as improving productivity. « Migrating to Google Cloud has given us the flexibility and control we needed, as we’re able to choose on-demand billing where that works best for us, » says Jonathan. « Since migrating, we’ve reduced our costs by 50%. » The granularity offered by Google Kubernetes Engine in resources management and scaling reduced the costs of compute instances. With most of the workloads now running on Google Kubernetes Engine, Gamned saved a lot of money thanks to the committed-use discounts.

Gamned’s IT team is now planning to migrate its machine learning projects to Dataproc, which will enable its ad traders to better optimize ad campaigns automatically. It also plans to start using Cloud Bigtable for its dynamic banner generation tool, which receives multiple requests per second and requires a high performance database.

« We need to keep innovating, both in terms of the technologies we are working with, and the products that we offer, » says Jonathan. « With Google Cloud, we have the capacity and tools to continue adapting our infrastructure and finding new ways to better serve our clients. »

gamned-logo

A Propos

With more than 140 experts working across six territories, Gamned creates targeted multichannel programmatic ad campaigns for brands and agencies by combining human expertise and cutting-edge technology.

Location: France


INDUSTRIE
  •  Technology

TECHNOLOGIES

  • Google Cloud
  • Kubernetes Engine

PARTENAIRE

Retrouvez ce témoignage chez notre partenaire Google Cloud



oiseau-bleu-svgAsset 3

Vous avez un projet ?

Nos équipes répondent à vos questions 

Programmer un call
TRANSFORMATION CLOUD

Découvrez nos réussites
de transformation numérique

Modernisez-vos-applications

Modernisez ses applications

Devenez Cloud Native et passez à l'échelle avec Google Kubernetes Engine.

moderniser-vos-applications

Moderniser son infrastructure

Sécurisez et fiabilisez votre plateforme avec Google Compute Engine et Kubernetes Engine

moderniser-votre-plateforme-data

Moderniser sa plateforme Data

Moderniser votre Data Lake et votre Data Warehouse avec BiqQuery, Looker, Dataflow & Looker

extraire-le-potentiel-de-ses-données

Extraire le potentiel de ses données

Extraire les insights pour faciliter les choix stratégiques et monétiser la donnée (data sharing, dataViz)

DECOUVRIR NOS PUBLICATIONS TECHNIQUES

Dernières publications

L'impact de l'automatisation sur la gestion des ressources cloud

Le Développement Durable dans le Cloud

Google Cloud Summit France '24