Customer story · Retail technology
Otter: from ecommerce infrastructure to data and forecasting.
CloudHero helped Otter modernize its Magento ecommerce infrastructure on AWS and Kubernetes. Later work covered a migration to Google Cloud, a BigQuery data platform and inventory forecasting, followed by operational stabilization and transfer to Otter’s teams.

- Earlier platform
- Magento, AWS and Kubernetes
- Later delivery
- GCP, BigQuery and forecasting
- Operating model
- Build, operate and transfer
The starting point · 2019–2020
Four storefronts needed a more consistent way to ship software.
The original Otter project covered four Magento storefronts associated with Otter, Tezyo, Salamander and Aldo. Their implementations differed, and the infrastructure relied on scripts that made daily operations difficult to maintain.
Development and testing shared a single pre-production environment. Creating another environment took time, production issues were difficult to investigate, and onboarding developers was cumbersome. Otter wanted a platform that could support both ecommerce growth and a larger engineering team.
“We knew that to become agile and have the capability to scale both our engineering team and our eCommerce business, we needed to start with a great foundation.”
Ecommerce engineering
Package the applications and make their dependencies explicit.
CloudHero containerized the applications, packaged them with Helm and deployed them through CI/CD pipelines. Kubernetes provided a common deployment model across environments. The original customer account describes more autonomous releases, more consistent environments and less operational complexity.
The work also addressed the state that Magento needed outside an individual application container. Otter used Aurora MySQL for its database, ElastiCache Redis for caching and sessions, and Amazon EFS for shared persistent files. Amazon SQS supported queues for new services.
| Need | Implementation in the case | Why it mattered |
|---|---|---|
| Repeatable releases | Containers, Kubernetes, Helm and CI/CD. | Give the storefronts a common packaging and deployment process. |
| Database and sessions | Aurora MySQL and ElastiCache Redis. | Support Magento’s database, cache and session requirements across application instances. |
| Shared files and queues | Amazon EFS and SQS. | Provide shared persistence and asynchronous communication where the applications needed them. |
| Traffic and diagnosis | Istio, ELK, Prometheus/Grafana and New Relic. | Support traffic routing, logs, monitoring and application performance investigation. |
Istio also enabled the team to serve different store versions to user segments for marketing experiments. During the COVID-19 closure of physical stores, Otter reported that the cloud platform helped it respond to increased ecommerce demand.
Historical source: the original CloudHero customer interview, published in May 2020. It describes the earlier AWS engagement; the later Google Cloud work is documented below.
Watch the original Otter engineering conversation on YouTube ↗
Later work · Customer recommendation dated 16 February 2026
Extend modernization into data access and inventory planning.
Otter’s later work with CloudHero went beyond the earlier AWS ecommerce platform. In a written recommendation, IT Manager Iulian Aldea confirmed a Magento migration to Google Cloud Platform, a data lake on BigQuery and an inventory forecasting system.
The recommendation describes CloudHero leading the build and implementation, operating and stabilizing the delivered components, and completing a structured transfer to Otter’s internal teams.
A data platform teams could use
Otter reported improved access to reliable data through the BigQuery platform. Bringing data engineering into the engagement connected infrastructure delivery with the information available to the business.
Forecasting with an operational handover
Otter reported that the inventory forecasting capability supported better planning decisions. The work included stabilization and transfer, giving internal teams a path to take ownership of the delivered components.
“CloudHero led the build and implementation phases, operated and stabilized the delivered components with clear routines, and completed a structured transfer to our internal teams.”
The later delivery is documented separately from the 2019–2020 AWS project. The customer recommendation describes the business capability and handover; it does not specify a forecasting model or accuracy metric.
For retail engineering teams
Connect the platform, the data and the team that will own them.
Otter’s experience shows two distinct modernization needs: a consistent way to run ecommerce applications, and a dependable way to use business data. A project can address either need on its own, but the delivery plan should account for the interfaces and responsibilities between them.
- Map shared application state. Identify databases, sessions, files and queues before changing how a storefront is deployed.
- Design for developer workflow. Separate development and testing needs, and make environment creation repeatable.
- Define the forecasting decision. Agree which inventory decision the output supports, the data it depends on and how usefulness will be evaluated.
- Include transfer in the scope. Decide who will operate the platform and data workloads after delivery, with documentation and a stabilization period.
For a similar starting point, explore data engineering and analytics, Kubernetes platform engineering or an ecommerce peak-readiness assessment.