Agile Order Entry Migration to AWS
This Global footwear and sports apparel giant processes over $13B of planned inventory orders that culminate during quarterly ordering cycles. Within these order deadlines, an array of product merchants (ranging from large retailers to mom-and-pop sports stores) place delivery requests for the next season. This requires an order fulfillment system capable of handling the rush of all clients placing orders within a week, four times per year.
The previous dedicated solution used expensive legacy hardware (a 20th Century behemoth) for management. The requirement was to migrate the entire order entry chain from the dedicated solution to a dynamically scalable cloud solution using Amazon Web Services.
STATEMENT has been a partner of this Fortune 500 retail company’s eCommerce team for years and knows the business logic and software systems. Since numerous Amazon Web Service implementations have been completed, STATEMENT engineers were brought onto this project immediately.
The adopted philosophy of the migration was “lift and shift.” Each component of the application was duplicated within the Amazon environment using as close to the original physical hardware footprint as available. Migration included over a hundred nodes consisting of Java application servers, Oracle 11i, MongoDB, Apache Web, Apache SOLR, RabbitMQ, and various other technologies.
At the start of the project, the timeframe for the migration was 12 months. Through use of Agile Scrum project methodologies, STATEMENT was able to accelerate the timeframe to 9 months. Further, velocity increases changed the timeframe to 7 months, and finally to 5 months after only 2 months of development. The project was a raving success.
The fortune 500 retailer is saving millions of dollars a year in hosting costs from the migration. The eCommerce group can be more nimble with software releases as the entire development pipeline has been moved to AWS. The continuous integration pipeline exists within AWS itself.
Phase 2 has begun which will allow the ability to scale computing resources up as order deadlines approach, then scale them back after the deadline. Hosting costs are estimated to be 20% of the original yearly expense.