Flower retailer bursts into bloom with Aiven for Apache Kafka® at the heart of its cloud data infrastructure
Aiven makes setting up cloud databases so simple anyone can do it. Our set-it-and-forget-it solutions take the pain out of cloud data infrastructure.
Leading flower retailer BLUME2000 SE needed to rebuild the legacy system underpinning its online business to improve the buying experience for its customers and enable scalability. As part of the refresh, it selected Aiven for Apache Kafka® running in Google Cloud to ensure the rapid transfer of data between applications. With the new responsive and scalable system in place, BLUME2000 SE was able to double its online business during the COVID-19 pandemic. Since then, the robust cloud data infrastructure has enabled the retailer to introduce new services and functionalities bringing new levels of customer engagement.
Established in 1974, BLUME2000 SE is a flower shop chain that has grown into Germany’s leading flowers and plants retailer. In addition to its 230 stores, it has a significant online business which has been running for 20 years. Competition has increased in recent years as all the major supermarkets now sell flowers in-store and online, and online-only florists have entered the market, many offering subscription services. To maintain its leadership position, BLUME2000 SE differentiates itself by providing the best possible experience to its customers, and aims to make that experience seamless across its in-store and online businesses.
To fulfil its customer-centric strategy, BLUME2000 SE needed to refresh the legacy infrastructure and software that underpinned its online business. Developed in-house over many years by many different people, its structure had become complex and integration with other systems or new technologies was difficult. “The legacy system wasn’t keeping pace with the business — it was hindering the buying experience for the customer and the efficiency of purchasing products for us. It was time consuming to add new features and functionality,” says Philipp Westphalen, FullStack Software Engineer, BLUME2000 SE.
Westphalen’s team undertook a thorough assessment of the legacy system, created a system modernization strategy, then set about choosing the optimal technology stack. The new system was designed with self-contained systems (SCSs) communicating asynchronously to each other, ensuring scalability as the online business grew. The team also took into consideration that it would operate in a hybrid cloud environment. BLUME2000 SE had taken the strategic decision to migrate as many workloads as possible to Google Cloud, but some data and apps would remain on-premises primarily for compliance and regulatory reasons.
Apache Kafka was chosen as the best open source platform to enable near real-time communication between the SCSs and other systems, enabling data streaming at scale on Google Cloud, the public cloud of choice for BLUME2000 SE.
For Westphalen’s team of developers, with limited operational knowledge, managing the Kafka clusters became challenging. “Managing Kafka is complex. Doing it by yourself is really hard,” says Westphalen. “As soon as we started using Kafka for bigger loads, we ran into trouble. We had limited resources yet we were spending a lot of time managing and fixing. We had to scale up our on-premises solution because we destroyed our clusters several times.”
The team investigated Apache Kafka as a fully managed service and selected Aiven as the best vendor. Important criteria included the ability of BLUME2000 SE to host the Kafka clusters in Google Cloud as well as providing a Terraform interface allowing the team to orchestrate the Aiven service alongside the rest of the technology stack. The transparent, all-inclusive pricing and easy scalability of Kafka clusters were also key benefits.
BLUME2000 SE seamlessly connected Aiven for Apache Kafka with the rest of the infrastructure underpinning the online shop. Data from nearly every transaction between services now runs through Kafka. “If a customer indicates they want to buy an item, it is written into Kafka and that data is then used by several systems such as the shipping, order history and newsletter systems,” says Westphalen. “It’s a lot of important data and it’s critical that data moves around fast and reliably.”
A key part of the refreshed BLUME2000 SE infrastructure is the Google BigQuery data warehouse. Apache Kafka MirrorMaker is used to replicate required data from the on-premises Kafka clusters to Aiven for Apache Kafka clusters in Google Cloud for ingestion into Google BigQuery. “We couldn’t have chosen Aiven if we couldn’t deploy in Google Cloud. Eventually, all our data infrastructure will be there,” says Westphalen.
Westphalen and his team no longer spend time maintaining the Kafka clusters. Aiven provides 99.99% uptime and ensures near zero downtime during maintenance operations like security patches and version upgrades. “Kafka is an important part of our data infrastructure and essential to the smooth running of the remodelled online shop,” says Westphalen. “Aiven takes care of everything — provisioning, configuration, upgrades. Kafka just works. It’s such a relief!”
The timing of the system refresh was fortuitous as it was completed just before the major COVID-19 lockdowns occurred in Germany. BLUME2000 SE closed its shops and its customers went online. During that period, online revenue for the retailer almost doubled so the company had a strong year despite the challenging conditions. If the legacy system had been in place, BLUME2000 SE wouldn’t have been able to handle the increase in customers, transactions and data, putting the company’s profitability at risk.
As the COVID-19 restrictions were lifted and customers returned to brick and mortar shops, day-to-day online sales at BLUME2000 SE reduced but remained significantly higher than previously. Valentine’s Day and Mother’s Day are two of the most significant days of the year for BLUME2000 SE and an opportunity for Westphalen’s team to see the updated system operating at peak load. Despite the volume of order-related transactions increasing by 400% – in part due to new events – the high performance and scalability of Aiven for Apache Kafka ensures the data moves rapidly between the services.
“We used to have issues with loading times for our customers. Today, with Aiven for Apache Kafka, we have very low latency which has transformed the customer experience. Even on occasions like Mother’s Day when we have more customers, more transactions and more data, everything operates seamlessly,” says Westphalen.
“I’m pleased with the team’s decision to use Aiven for Apache Kafka. Our developers can focus on innovation and delivering an awesome experience to our customers,” says Jonas Clusen, CIO at BLUME2000 SE.
The retailer has a goal to migrate everything from its on-premises environment to Google Cloud, and Aiven for Apache Kafka will grow as the company’s cloud data infrastructure expands. As part of this growth, Westphalen’s team is exploring Aiven for PostgreSQL® to meet its high performance, relational database needs.
In tandem, Westphalen’s team is collaborating with the IT department that supports the in-store business, and the ongoing integration is already bringing success. For example, BLUME2000 SE recently launched a customer loyalty program that operates in-store as well as online, and introduced a new service to deliver flowers ordered in-store to customers’ homes using Wolt and Uber delivery services across different cities in Germany.
“It’s an exciting time at BLUME2000 SE and it's gratifying to see customers respond positively to the new services my team has helped develop,” says Westphalen.
