For eBay's 3-year modernization, attention turns to developing, using open source infrastructure
- EBay is stepping into the hardware market with custom-designed servers that will be publicly available by the end of the year through open source, according to an announcement by VP of Platform Engineering Mazen Rawashdeh on Wednesday. Open sourcing infrastructure is part of eBay's three-year replatforming and modernization efforts.
- Combing through each layer of its technology stack, eBay replatformed without incurring incremental costs. The company built and deployed its custom hardware prototype in nine months, reducing dependence on third parties and homogenizing infrastructure, while renewing commitments to the open source philosophy throughout its development and relationship with the open source developer community.
- The company has already designed new hardware and an AI-engine in-house; adopted edge computing architecture and more open source; and decentralized its data centers through a Point of Presence strategy. These efforts have improved flexibility, predictability and control for eBay, which has daily data footprint of 500 petabytes, 175 million active users and 300 billion data queries a day, according to Rawashdeh.
Rawashdeh credited the company's new AI engine for leveraging advanced capabilities at scale, including computer vision and image search.
EBay strengthened its AI chops earlier this year with a new VP and chief scientist of AI, Jan Pederson, who joined the company from Twitter with a mission to focus on computing vision, dynamic pricing and search and personalization functions. The company has worked to embed AI capabilities throughout its customer experience and backend processes.
EBay's experience reinforces the importance of culture to digital transformation. The company has instilled an learning atmosphere where creativity is encouraged and risk aversion is counteracted, because "with the right culture in place, the technology and innovation will follow," according to Rawashdeh.
New computing architectures and work management processes can afford little change if the IT workforce isn't buying in to a new scheme of handling data and or understanding how the business relates to others internally and externally. With millions of dollars on the line each year for digital transformation efforts, not addressing the human component is a costly mistake.
For some companies, such as Spotify, getting IT behind a new system means steadily migrating individual teams to a new system; other companies have opted for more sweeping, uniform changes.
Shifting to open source technologies and development within a company can facilitate development and sharing across teams to avoid redundant code writing, among other benefits. But the transition can also prove disruptive.
When C.H. Robinson shifted to open source technologies, more autonomy was given to development teams, changing traditional team structures. As a result, the speed of deployments rocketed 162% per week, at times moving the organization too quickly. With time and adjustment though, disruption was mitigated, bringing the company from the status quo to a digital culture and system that better fit its future needs.
Follow Alex Hickey on Twitter