Kose moved its K-PAD sales tablet system to AWS containers — failures fell to zero
The cosmetics maker rebuilt K-PAD on ECS and Fargate in October 2021: login incidents went to zero and processing speed rose sixfold.
What was changed
K-PAD is Kose's tablet-based sales support system: beauty consultants on department store floors use it to pull up customer and product information while serving shoppers. Server-side, the whole thing ran on a single virtual machine on AWS — and malfunctions were frequent. In October 2021 Kose migrated the system onto a container platform, provisioning multiple containers on Amazon Elastic Container Service with the serverless Fargate option, eliminating the need to build and manage VM or container execution environments directly.
The infrastructure change delivered immediate, countable results: login failures, previously recurring, went to zero, and processing speed improved sixfold. Kose is now optimizing the application with microservices in view, so that future feature additions and repairs shrink the blast radius of any change.
The revamp doubled as a capability build. Kose constructed an in-house development setup that pulls in its development vendor and AWS Professional Services consultants, standardizing as it goes and accumulating development know-how inside the company rather than leaving it with contractors — a deliberate shift toward owning the system it runs on the shop floor.
Why it worked
A single virtual machine meant every department-store consultation depended on one failure-prone box; containers spread the load and the risk.
Fargate's serverless model removed environment management work, speeding development and cutting cost at the same time.
Zero login failures and 6x processing came from infrastructure redundancy, not an application rewrite — the cheapest fixes are often under the app.
Embedding vendor and AWS consultants inside an in-house team turned a one-off migration into lasting internal capability.
What can be applied
One VM is a single point of failure for customer-facing systems: containers plus serverless buy redundancy and speed without a rewrite.
Aftermath
Kose continues developing K-PAD on the container platform, with application optimization underway and microservices decomposition as the stated next step to further shrink the impact of future feature additions and fixes.