How do you reduce the cost and complexity of an online marketplace’s search function?
By using the Elastic Stack to scale search and simplify data access.
Kaidee’s Journey with Elastic
Kaidee is Thailand’s largest online marketplace for second-hand goods. Its platform has over 650,000 daily visitors and 40 million daily page views. This year, it aims to double the number of daily visitors. These growth plans increased the focus on the scalability of the platform. Kaidee had been relying on an expensive, proprietary appliance solution to power user searches across the ads posted by its sellers. It was, however, inflexible to scale to Kaidee’s growth plan.
Elasticsearch provided an attractive alternative because as an open-source solution it was less expensive but equally powerful and highly scalable. Kaidee was also familiar with Elasticsearch as it was used by the infrastructure team to support browsing. In addition, Kaidee saw an opportunity to simplify access to data and analytics with Kibana. With an easy to use interface, the team could quickly set up visualisations and explore data on its own.
Mark Hollow, CTO at Kaidee said, “Kaidee was started because we believe that buying and selling second-hand items is useful and should be accessible to all Thai people. We’re building the best team ever to make that happen.”
Kaidee initially used Elasticsearch for six months alongside its appliance solution to benchmark and compare performance. When it came to text search, Kaidee found Elasticsearch coped with the complexities of searching through the Thai language. This added performance, combined with a significantly lower cost of ownership, made Elasticsearch the clear winner.
Kaidee switched to its appliance solution, making Elasticsearch the primary search engine for its platform. It lets users search for products and services in their area using Thai. The posted ads are initially submitted to the platform via web or mobile apps. They are then sent into an open-source database and extracted into Elasticsearch for simplified user search.
The user experience is fast and seamless with users required to enter in just a few key words for useful and accurate results.
Kaidee has also resolved the issue of scale. The infrastructure team can increase capacity vertically or horizontally as needed, and with Elasticsearch easier to manage than the previous solution, they are freed up to focus on new products and features.
Kibana has opened up analytics capability to the entire business. With live metrics presented in a simple and easy to understand manner, it gives users quick insights into the business and supports improved reporting. The analytics team can work more efficiently to drive the business forward rather than getting involved in every data request.
Kaidee is also using Elasticsearch and Kibana to collect and visualise information from application logs. This, combined with the use of the X-Pack Security, supports Kaidee to protect the performance and data behind its platform. For instance, it no longer needs to worry about unauthorised access to data and external threats are also minimised.
Kaidee is now also evaluating Graph from Elastic with the aim to make it easier for the analytics team to extract new insights from its data.
Mark concluded that Kaidee not only benefited from Elastic’s technology but also its support.
Elastic has been very responsive and we feel we can ask them anything. Even on the development side of things, they’ll offer us their suggestions and support. It’s been great being part of the community and having the opportunity to share our key learnings with others at Kaidee’s and Elastic’s events. We look forward to further collaboration in the future.