- Популярные видео
- Авто
- Видео-блоги
- ДТП, аварии
- Для маленьких
- Еда, напитки
- Животные
- Закон и право
- Знаменитости
- Игры
- Искусство
- Комедии
- Красота, мода
- Кулинария, рецепты
- Люди
- Мото
- Музыка
- Мультфильмы
- Наука, технологии
- Новости
- Образование
- Политика
- Праздники
- Приколы
- Природа
- Происшествия
- Путешествия
- Развлечения
- Ржач
- Семья
- Сериалы
- Спорт
- Стиль жизни
- ТВ передачи
- Танцы
- Технологии
- Товары
- Ужасы
- Фильмы
- Шоу-бизнес
- Юмор
KCD Denmark 2024: High Performance Data Backend for Your AI/ML with MinIO and Cilium Cluster Mesh
Session: High Performance Data Backend for Your AI/ML with MinIO and Cilium Cluster Mesh - George Zubrienko & Vitalii Savitskii
Owning full value chain of your product opens a vast array of possibilities to optimize operations and generate more revenue. However, utilizing commercially available tools in this area greatly diminishes the net result and doesn’t scale well with problem complexity. Thus, In ECCO we rely on highly complex demand forecasting powered by open-source LightGBM framework and employ best-in-class Mixed Integer Optimization framework Ortools from Google to fulfill online orders and replenish our stores, to have the right product in the right time, in the right place. Still, all these marvels are of little use if your data takes ages to load, or if your input datasets take hours to produce. In this talk we’ll demonstrate how embedding your storage into a cluster mesh with your Spark and ML containers via Cilium can outperform classic commercial blob storage offerings, as well showcase the reference architecture that we currently run in production. We'll also provide sample code to provision a pair of clusters (AKS/EKS) operating a shared service through Cilium Cluster Mesh
Видео KCD Denmark 2024: High Performance Data Backend for Your AI/ML with MinIO and Cilium Cluster Mesh канала Cloud Native Nordics
Owning full value chain of your product opens a vast array of possibilities to optimize operations and generate more revenue. However, utilizing commercially available tools in this area greatly diminishes the net result and doesn’t scale well with problem complexity. Thus, In ECCO we rely on highly complex demand forecasting powered by open-source LightGBM framework and employ best-in-class Mixed Integer Optimization framework Ortools from Google to fulfill online orders and replenish our stores, to have the right product in the right time, in the right place. Still, all these marvels are of little use if your data takes ages to load, or if your input datasets take hours to produce. In this talk we’ll demonstrate how embedding your storage into a cluster mesh with your Spark and ML containers via Cilium can outperform classic commercial blob storage offerings, as well showcase the reference architecture that we currently run in production. We'll also provide sample code to provision a pair of clusters (AKS/EKS) operating a shared service through Cilium Cluster Mesh
Видео KCD Denmark 2024: High Performance Data Backend for Your AI/ML with MinIO and Cilium Cluster Mesh канала Cloud Native Nordics
Комментарии отсутствуют
Информация о видео
14 декабря 2024 г. 15:56:07
00:34:53
Другие видео канала
