eBay has embarked on a large-scale reconfiguration of its architecture: designing custom hardware and a dedicated artificial intelligence engine, decentralizing the data center cluster, evolving to cutting-edge computing, and developing open source technology solutions.
In the process of completion, as the new servers are already operational; their architecture will be made public, in fact becoming open source. Committed for three years to a project to renew its platforms and modernize its backend infrastructures, eBay announces that it will build its own server designs and offer them in open source by the end of 2018.
Launched by Facebook 7 years ago, the Open Compute Project (OCP) is an initiative to share server designs and make them available in open source.
The latter has grown over the years with the support of leading IT companies such as Apple, Microsoft, Google, HPE, Rackspace and Cisco.
However, some heavyweights are missing, such as eBay, which announced last weekend its intention to develop its own servers and share its open source design so that other companies can use them for their needs. If the U.S. online retail giant has not made any announcements regarding its OCP membership by now, it is very likely that it will end up joining in the coming months.
“As part of an ambitious three years of effort to reconfigure and modernize our back-end infrastructure, eBay announces its own custom servers designed by eBay for eBay. We plan to make eBay servers publicly available through open source in the fourth quarter of this year,” the company said in a post. “The reconfiguration of our core infrastructure included the design of our own hardware and IA engine, the decentralization of our data center cluster, the adoption of a next-generation IT architecture and the use of the latest open source technologies.
Leveraging IA on a Large Scale
Among the technological bricks used by eBay are Kubernetes, Envoy, MongoDB, Docker, and Apache Kafka.
The infrastructure developed by the e-merchant allows it to process 300,000 million daily requests for a data volume of more than 500 petabytes.
“With the transformation, we’ve achieved and the large data flowing through eBay, we’ve used open source to create an internal AI engine that is highly shared among all of our teams and aims to increase productivity, collaboration, and training. It allows our data scientists and engineers to experiment, create products and customer experiences, and leverage IA on a large scale,” eBay said.