We provide you with
EasyStack AI Infrastructure Suite
Designed to build AI-ready infrastructure that is more integrated and delivers higher performance for AI applications.
AI-Ready Infrastructure Supporting AI Models and Modern Application Development
Built on ECF and ECNF infrastructure products, the EAI Infrastructure Suite collaborates with partners to create AI-ready infrastructure. This optimized infrastructure product provides support for private AI models and AI applications, accommodating a broader range of open-source frameworks and AI solutions.
Why Choose EasyStack AI Infrastructure Suite?
GPU, NPU, DCU Support
The EasyStack AI Infrastructure Suite leverages the EOS digital native engine and the ESCL Cloud Operating System to provide unified management of heterogeneous GPU, NPU, and DCU device resources.
EAI Suite Overall Activation
The overall activation of the EAI Suite involves leveraging the EOS digital native engine and the ESCL Cloud Operating System to provide unified management of heterogeneous GPU, NPU, and DCU device resources
Compute Pooling
EasyStack natively supports batch scheduling frameworks, making it easy for you to build an AI compute pooling infrastructure.
Evolvable Infrastructure
Supports a smooth transition from any deployment form of cloud infrastructure or cloud-native infrastructure to AI-ready infrastructure. This ensures consistent management, permissions, monitoring, logging, alerting, and access across all stages of evolution.
Comprehensive acceleration of computing networks, empowering cloud-native AI infrastructure.
Unified planning, stable isolation, and on-demand deployment for AI, big data, and web applications.
Unified network model for bare metal containers, bare metal, and instances, with unified segmentation and on-demand interconnection, ensuring computing power isolation.
Single-layer container network with multiple direct pass-through modes for performance acceleration.
Using bare metal container direct pass-through networks transforms two-layer networks into a single layer, enhancing application inter-access performance without loss.
Unified load balancing, with cloud-to-cloud and external traffic flowing on demand.
External traffic directly passes through to containers and cloud hosts, ensuring load isolation for multi-tenant traffic.
AI computing infrastructure empowers modern application and industry scenario innovation.
Integrated, complete, high-performance AI computing, network, and storage infrastructure products
Large-scale bare metal containers, bare metal, instances, and instance containers all support AI computing specifications. High-speed AI cloud networks support InfiniBand and RoCE, enhancing access performance for computing and storage. High-performance, low-latency cloud storage supports NVMe-oF and RDMA to meet the demands of large-scale data access.
Intelligent computing resource scheduling management, refined resource management, and operational support
The AI infrastructure is responsible for scheduling computing resources (CPU, GPU, NPU, DCU, etc) and utilizes Volcano for AI task scheduling, supporting communication frameworks such as NVLink and HCCS. It features comprehensive and unified resource monitoring and management, reducing the overall TCO of the AI infrastructure by over 50%.
Large-Scale, Safe, and Reliable AI Infrastructure Solutions
With 10 years of production experience involving 100,000 nodes, our infrastructure supports single clusters of over 10,000 nodes at scale. Bare metal containers offer a secure and reliable compute option with multi-tenant support, performance isolation, and fault isolation. Coupled with a continuous and smooth upgrade capability in a production environment of 20,000 nodes over 5 years, this makes AI infrastructure construction more stable and secure.
Support for multiple AI hardware Segmentation and Unified Management of Multiple Compute Pools
Not only does it support segmentation and management of NVIDIA GPUs, but it also supports segmentation and management of Ascend NPUs and DCUs. It provides heterogeneous isolation and unified management capabilities for general compute and AI compute resources across multiple tenants and clusters, ensuring that there is no interference between tenants' resources, data, and applications while enabling full sharing and reuse.
Support for an Open Ecosystem of Scheduling, Training, and Inference Frameworks
The infrastructure provides standardized APIs and application template management based on cloud-native standards. High-performance storage and networking meet the deployment requirements of various distributed high-performance middleware, databases, and big data, supporting open AI frameworks as well as training and inference platforms.
Learn more about EasyStack AI Infrastructure Suite
For more inquiries about use cases and technical architecture, please contact our sales and technical support team.