Case Study

DoiT helps Onedot fuel ecommerce growth with efficient AWS architecture

E-Commerce, Internet
Amazon Elastic Kubernetes Services, AWS, AWS EC2

Meet Onedot

One-third of the world’s population now shops online, accounting for over 20% of all retail sales annually. For every one of those billions of physical products sold, a corresponding set of product data is needed to describe it accurately in its digital shop or marketplace, so that shoppers can find what they are looking for. The trouble is, every supplier describes its products in a different way. Wading through this sea of data can be a major operation for retailers, draining resources and leading to poorly classified products on their sites.

That’s why so many leading international retailers turn to Onedot. With its AI-powered platform, Onedot automates the entire process of sourcing, onboarding, and syndicating product data for every product sold online. Over 500 million products have already been processed by the Onedot AI, helping retailers offer their customers a better shopping experience, while hugely reducing their overheads.

The challenge

With a modern cloud-first architecture hosted on AWS, Onedot uses Apache Spark on Amazon EMR as its core data-processing engine. But with such a vast amount of data passing through its pipelines, Onedot wanted to make sure that it was using its compute resources in the most efficient way possible, automatically scaling usage up and down as needed, in order to optimize its cloud spend. At the same time, it was important to avoid an increase in costs. “We were having some issues with cost management with our AWS workloads,” explains Tobias Widmer, Founder and CTO of Onedot. “We turned to DoiT for help with resource planning and general architectural guidance, so that we could get the most from our data processing on AWS, while keeping our cloud bill down.”

The solution

Onedot began working with DoiT in the spring of 2023 by explaining its current architecture and challenges.

DoiT then embarked on a two-pronged service approach. For one, DoiT took Onedot through an AWS Well-Architected Review to identify opportunities to align Onedot’s workloads and architecture to AWS best practices.

The second part of the service engagement focused on optimizing resource usage to reduce cloud costs. These optimization opportunities centered around three key areas:

Automating AWS instance capacity with Flexsave
The first step Onedot took to optimize its cloud spend was to implement DoiT Flexsave, automating the management of its AWS Savings Plans coverage to continually make the most of its AWS commitment-based discounts. This gives Onedot the freedom to increase or decrease its cloud usage as needed, without incurring huge bills, or having to spend time continually adjusting its commitments.

Continually refining infrastructure
Onedot has also begun using the DoiT console for reporting on its AWS cloud usage, working with DoiT consultants to set up automated alerts for unexpected patterns of usage that are likely to incur higher costs. Armed with the insights provided by the DoiT console, Widmer and his team then worked with the DoiT consultants to analyze certain spending, and how they might refine Onedot’s AWS infrastructure to reduce costs.

Getting the big decisions right
A key discussion surfaced from Onedot’s use of Apache Spark on EMR, investigating whether Onedot could reduce its cloud expenditure by managing its own Kubernetes cluster and running Spark on that instead. Working with the DoiT team, Onedot examined in depth the relative costs of the various system permutations, and ultimately decided that using EKS – the managed Kubernetes service from AWS – would be more cost effective than either EMR or running its own Kubernetes cluster.

“This whole balancing of costs versus savings, and the pros and cons of various infrastructure, is the kind of thing that DoiT helps us with,” says Widmer. “We’re not beginners with AWS, so we’re definitely challenging DoiT with our requirements, and they help us come up with advanced solutions on an architectural level.”

The result

Reducing AWS cloud expenditure by more than 20%
Having only been working together for three months, Onedot is still in the early stages of its journey with DoiT, but by automating its AWS workload commitments with Flexsave, and working with the DoiT team to rebalance its architecture, Onedot has already reduced its AWS bill.

Facing the future as a scalable, leaner, more efficient operation
Onedot is now working with DoiT to strengthen its security posture with the goal of acquiring international certifications for its Information Security Management System. Meanwhile, Onedot and DoiT continue to work together to refine Onedot’s architecture to further improve efficiencies, ensuring that it can offer a more stable compute environment, capable of delivering product data to its customers more quickly and reliably.

What's next?

“As we continue to work with DoiT, they help us to stay fit and lean,” says Widmer, “to make sure that as our company grows, our cloud costs don’t increase at the same rate. As our partner, we know DoiT will help us to keep finding the right way forward, to choose the right option to keep growing our business and delivering the product data our customers depend on.”

Tobias Widmer, Founder and CTO
“This whole balancing of costs versus savings, and the pros and cons of various infrastructure, is the kind of thing that DoiT helps us with. We’re not beginners with AWS, so we’re definitely challenging DoiT with our requirements, and they help us come up with advanced solutions on an architectural level.”

Learn more about how DoiT can help you

Latest case studies