AtTradera, we have a saying, ‘Don’t be square, get circular.’ Our online circular marketplace only allows used, crafted, or restored items and spare parts for repairs to be sold. Our ethos to reduce and re-use is engrained throughout our business operations, down to our IT infrastructure.
As one of the largest re-commerce platforms in Sweden, Tradera welcomes six million website visitors per week and features five million item listings at any time. We help useful merchandise change hands every three seconds and our systems process 300 bids per second at peak times. Sustainability sits at the center of our corporate mission, and influences our IT provisioning and management decisions.
Growing into scalable cloud infrastructure
Shoppers across Europe have embraced our model for sustainable shopping for everything from clothing to sports equipment. By 2018, we outgrew our on-premises data center. We also needed a fresh approach to our content delivery network (CDN) which was neither fast nor robust enough. Every week, we worried about whether our systems would be able to handle the surging traffic for the upcoming week.
We evaluated multiple cloud providers and concluded that Google Cloud best fit our needs regarding performance, cost effectiveness, and managed services, and the Kubernetes support weighed heavily.
We saw immediate improvements with a lift and shift of our production environment to a managedGoogle Kubernetes Engine (GKE) environment onGoogle Cloud. In the process,Compute Engine andCloud Storage proved useful, kicking off our evolution to a more modern, service-oriented architecture using Kubernetes. For content caching, we chose to use Cloud CDN to get the speed and scalability that are vital to our success.
It wasn’t long before we realized our existing data warehouse wasn’t serving us well either. Queries were slow. We were always thinking about whether we had space for new data. Expanding disks would mean rebuilding the whole server. We wanted to integrate withAnalytics 360, but bringing in any external data set was quite painful.
That’s when we contacted Google Cloud managed services providerDevoteam G Cloud to help us migrate our data warehouse toBigQuery. First, their team helped us with a pre-study to get a better sense of the migration project scope and how the ELT process would work. We were impressed by Devoteam G Cloud’s expertise, efficiency, and collaboration. They helped us complete the migration within a few months, which we would not have been able to do by ourselves.
We really like how the solution turned out. OurETL performance quadrupled with BigQuery andCloud Composer. Now we don’t have to choose which tables to bring in or worry about how long to store data. The user experience is good, the speed is great, and it’s very cost-effective.
Greater freedom to evolve our product offering
Without the constant concerns about our IT environment, our developers can more freely experiment with new features to improve Tradera buyer and seller experiences. For example, we added image searching withVision AI and Vision API Product Search to help shoppers find items on our website. Using machine learning, the service compares a given product photo, such as for a red dress, with our product listing images. We can then provide shoppers with visually and semantically similar item suggestions.
After experiencing bottlenecks in our previous environment, we deployed Elasticsearch to improve customer experiences with our website. Now, we have three Elasticsearch clusters that don’t cause us headaches, work great on Google Cloud, and automatically scale up and down on busy days so we don’t have to give them a second thought.
We’re much more sophisticated with our data today and can now effectively conduct analysis to improve user experiences. We have a team of data analysts that use BigQuery to harvest insights much faster thanks to powerful tools that help develop models and easily get them into production.
For example, we wanted to make selling easier, so we analyzed the time it takes sellers to post their ads for items for sale. Using BigQuery and Analytics 360, we created a model based on data from various sources, including our transactional data, to intelligently suggest item attributes to sellers while they’re filling out their product listing forms.
Continuous deployment agility
Our developers can deploy small service enhancements frequently since we don’t have to worry about a lack of compute resources, such as those needed to talk to services remotely. It’s so easy with GKE. Our operations team is also saving time thanks to the Horizontal Pod Autoscaler.
We’re glad we went with Google Cloud because it offers a lot of flexibility. Having access to such a great Kubernetes offering has been instrumental for us. You hear a lot of people saying that Kubernetes is complex, but we haven’t really felt that because of GKE. Administration is managed automatically, and we just specify the services, who they talk to, and how we want to scale them. It’s very easy to use.
Google Cloud has helped us shift from IT infrastructure management to innovation. If we want to expand to new countries and increase the size of our product catalog, we can do that without firefights previously caused by higher data volumes. In addition, we spend more time helping our business grow, aided by useful services likeTranslate AI and the Translation API to expand our reach in promising markets beyond Sweden.
We see a brighter future for our planet as more people realize the value of rethinking the items they buy and how they buy them. We are proud that Tradera isCarbon Neutral Certified and appreciate that Google Cloud aligns with our values,matching 100% of its electricity use with renewable energy and otherwise compensating for environmental impacts.
Learn more about Tradera’s sustainability efforts across all aspects of its operations.
Cloud BlogRead More