HomeTiger Migrates Ecommerce Platform to AWS with Zero Downtime with the Help of CloudHero

HomeTiger, which builds marketplaces that sell millions of home and garden products to customers across Germany and France, improved the reliability and scalability of its websites after it migrated a key part of its IT infrastructure to Amazon Web Services (AWS). The company decided to migrate after problems with its physical systems led to a costly 5-hour outage of its websites. With help from AWS Partner CloudHero, HomeTiger migrated its Elasticsearch database to AWS, using Amazon EC2 and Amazon VPC, and completed the project in less than 3 months with zero downtime.

Germany-based HomeTiger, founded in 2016, is a growing company that sells home and garden products through several online marketplaces serving customers in Germany and France. It plans to expand to other markets in Europe and beyond, but its IT infrastructure—a mix of physical and cloud services—wasn’t delivering the level of performance and reliability it needed. It had been considering migrating parts of its physical systems to a new cloud provider but accelerated those plans after an outage left customers unable to access HomeTiger websites for 5 hours.

The outage not only led to lost online sales, but also meant that HomeTiger’s online advertising campaign saw zero return on investment during the hours its sites were down. At the time, HomeTiger had been discussing migration options with AWS Partner CloudHero, which it had worked with previously on other projects. CloudHero suggested that, to start, HomeTiger should try migrating one part of its existing infrastructure to AWS—the Elasticsearch database that allows customers to search the company’s large database of products from multiple vendors.

Achieving that migration, though, involved managing a lot of complexity. HomeTiger sells around 12 million products from a variety of retailers and manufacturers, including Amazon, Home24, IKEA, Mayfair, and Otto Group. That means that HomeTiger must update its own database continually to keep up with changes—up to 12 changes every second—in product data from all of those sources. Because that database is critical to its business, HomeTiger had two key conditions for CloudHero: the migration had to be completed with zero downtime, and it had to be completed quickly.

HomeTiger decided to migrate Elasticsearch to AWS because it was looking for a cloud provider that could provide low latency for its product database and reliable connections to the many other services that the retailer uses. Scalability was also important to support HomeTiger’s roadmap for growth and expansion, and to satisfy its investors’ expectations around company performance. Having worked with CloudHero before, HomeTiger was confident that the DevOps specialist had the AWS experience and expertise needed to achieve its migration goals. “CloudHero was always helping us to find the best solutions,” says Guilherme Braga, chief technical officer (CTO) of HomeTiger. “And the decision to use AWS came from one of their suggestions about ways that we could perform better and improve a few key performance indicator (KPIs) that we were planning to focus on.”

CloudHero began working quickly, helping HomeTiger to win approval for project funding through the AWS Startup Migrate Program. Collaborating with HomeTiger, it also created a detailed migration plan that evaluated the retailer’s full development pipeline, assessed potential migration risks, and identified ways to make the most of the new ecosystem on AWS. The plan also spelled out strategies to rapidly roll HomeTiger’s infrastructure back to its physical systems if there were any problems with the new search capabilities on AWS. While CloudHero focused on managing the migration itself, it also provided the full-time services of a DevOps engineer to support HomeTiger’s IT team as the in-house engineers worked to improve other systems and optimize outcomes on AWS. “The biggest challenge we had during this process was around how we were going to migrate this data—12 million products—from A to B,” says Braga. “Because we have multiple websites in Germany and France, they needed to be fully operational and offering products to end users with no issues.”

After extensive testing and less than 3 months of preparation, HomeTiger and CloudHero were ready to migrate to a new setup on AWS using Amazon Elastic Compute Cloud (Amazon EC2), which provides secure and resizable compute capacity for virtually any workload, and Amazon Virtual Private Cloud (Amazon VPC), a way to define and launch AWS resources in a logically isolated virtual network. They chose to make the move at 0400 Berlin time when there’s typically low traffic on the retailer’s websites. They completed the migration within 30 minutes with no downtime. To ensure that online users wouldn’t notice any interruptions, HomeTiger temporarily suspended updates to its Elasticsearch data, instead directing searches through a snapshot of its database that CloudHero had transferred to Amazon Simple Storage Service (Amazon S3), which provides object storage built to retrieve any amount of data from anywhere. After going live with the migrated database on AWS, CloudHero continued to maintain two replicas of the Elasticsearch database on both AWS and HomeTiger’s physical systems and monitored database performance on AWS for 2 weeks before switching off the legacy database. “In the end, everything worked out amazingly,” Braga says. “It was really pretty much 100 percent done with no issues.”

With HomeTiger’s new infrastructure on AWS, the retailer’s websites are now much more stable, reliable, and scalable than before, and the company hasn’t experienced any more outages, Braga says. Before the migration, the company’s uptime was 91 percent—since then, it has increased to 95 percent.

Since the migration, the company has not experienced any issues with so-called soft downtime—when high latency caused long wait times for users loading HomeTiger’s websites, which happened often with its previous physical systems. Because its data is now managed through AWS servers in Frankfurt, Germany, latency issues are no longer a concern. Initial load times are now 10 percent faster than before, which not only makes for a better user experience but helps to improve the company’s visibility on search engines, Braga says.

Because of its success with the Elasticsearch migration, HomeTiger is now exploring several proofs of concept around other applications it might migrate to AWS. Its main focus is on marketplace services such as the database that handles its order management system, which will also need increased scalability as the company continues to grow. “We are very grateful to AWS—and to CloudHero for serving as a bridge between our company and AWS,” says Braga. “With the success of this project, it opens a lot of doors and possibilities for us moving forward.”