Blog
Coupang, a Fortune 200 technology company, manages a multi-cluster GPU architecture for their AI/ML model training. This architecture introduced significant challenges, including:
- Time-consuming data preparation and data copy/movement
- Difficulty utilizing GPU resources efficiently
- High and growing storage costs
- Excessive operational overhead maintaining storage for localized data silos
To resolve these challenges, Coupang’s AI platform team implemented a distributed caching system that automatically retrieves training data from their central data lake, improves data loading performance, unifies access paths for model developers, automates data lifecycle management, and extends easily across Kubernetes environments. The new distributed caching architecture has improved model training speed, reduced storage costs, increased GPU utilization across clusters, lowered operational overhead, enabled training workload portability, and delivered 40% better I/O performance compared to parallel file systems.
Suresh Kumar Veerapathiran and Anudeep Kumar, engineering leaders at Uptycs, recently shared their experience of evolving their data platform and analytics architecture to power analytics through a generative AI interface. In their post on Medium titled Cache Me If You Can: Building a Lightning-Fast Analytics Cache at Terabyte Scale, Veerapathiran and Kumar provide detailed insights into the challenges they faced (and how they solved them) scaling their analytics solution that collects and reports on terabytes of telemetry data per day as part of Uptycs Cloud-Native Application Protection Platform (CNAPP) solutions.
.png)
This article presents the collaboration of Alibaba, Alluxio, and Nanjing University in tackling the problem of Deep Learning model training in the cloud. Various performance bottlenecks are analyzed with detailed optimizations of each component in the architecture. Our goal was to reduce the cost and complexity of data access for Deep Learning training in a hybrid environment, which resulted in over 40% reduction in training time and cost.
This article describes how Alluxio can accelerate the training of deep learning models in a hybrid cloud environment when using Intel’s Analytics Zoo open source platform, powered by oneAPI. Details on the new architecture and workflow, as well as Alluxio’s performance benefits and benchmarks results will be discussed.
.jpeg)
Are you using SQL engines, such as Presto, to query existing Hive data warehouse and experiencing challenges including overloaded Hive Metastore with slow and unpredictable access, unoptimized data formats and layouts such as too many small files, or lack of influence over the existing Hive system and other Hive applications?
This article introduces Structured Data Management available in the latest Alluxio 2.2.0 release, a new effort to provide further benefits to SQL and structured data workloads using Alluxio.
.jpeg)
With this release comes the General Availability (GA) of Alluxio Structured Data Services (SDS), the subsystem of Alluxio responsible for managing and transforming structured data, such as databases, tables, and partitions.
TL;DR: First the news - Alluxio support for K8s Helm charts now available! K8s is a certified environment for Alluxio. Now the take away- Alluxio brings back data locality for the disaggregated analytics stack in K8s. How? Read on.
.jpeg)
We are delighted by the success of the inaugural Data Orchestration Summit on Nov. 7, 2019! Organized by Alluxio, this one-day event was sold out with nearly 400 attendees! Data engineers, cloud engineers, data scientists joined the talks of 24 industry leaders from all over the globe to share their experiences building cloud-native data and AI platforms. All session recordings and slides are now available.
