Published Jan 20, 2022

Ion Stoica — Spark, Ray, and Enterprise Open Source

Lukas Biewald engages with Ion Stoica, co-creator of Spark and Ray, delving into the art of founding companies, the influence of enterprise open source on innovation, and the intricacies of developing distributed computing frameworks.
Episode Highlights
Gradient Dissent - A Machine Learning Podcast logo

Popular Clips

Episode Highlights

  • Open Source

    Ion Stoica, co-creator of Spark and Ray, emphasizes the critical role of open source in the success of companies like Databricks. He explains that the decision to focus on open source was pivotal, as it was seen as essential for the company's credibility and success 1. This strategy was not without challenges, as the open-source model was not as common or proven at the time. Stoica recalls the early days when the open-source business model was a significant gamble, yet it laid the foundation for Databricks' eventual success 2.

    We decided that the success of the open source is necessary to the success of the company.

    ---

    The commitment to open source allowed Databricks to build a strong community and product, ultimately leading to widespread adoption and success.

       

    Monetization

    The monetization of open-source projects like TensorFlow and PyTorch presents unique challenges. Ion Stoica suggests that these frameworks, developed by large companies like Google and Facebook, are not primarily focused on direct monetization 3. Instead, they aim to enhance their cloud platforms, such as Google Cloud Platform, by optimizing these frameworks for their infrastructure. Stoica notes that hosted offerings are more valuable for distributed solutions, which is not the primary use case for these frameworks 3.

    Hosted offerings are more valuable when the solution is distributed, because then the value is to manage a cluster.

    ---

    This insight highlights the complexity of monetizing open-source projects, especially when they originate from large tech companies.

Related Episodes