Development Trend
Business Challenges
Business Challenges
Difficult in Predicting Needed Resources and High Costs of Self-Built Centers
The rendering services fluctuate drastically. When the services are idle, resources are wasted; during service peaks, resources become insufficient. In addition, constructing a local render farm requires a large number of IT resources, incurring high costs.The rendering services fluctuate drastically. When the services are idle, resources are wasted; during service peaks, resources become insufficient. In addition, constructing a local render farm requires a large number of IT resources, incurring high costs.Inefficient Rendering
A single rendering task takes dozens of hours on general-purpose servers. When the render farm is overloaded, the task will take much more time.A single rendering task takes dozens of hours on general-purpose servers. When the render farm is overloaded, the task will take much more time.
Various Storage Specifications
The servers in a render farm need to access the shared storage concurrently, so the shared storage bandwidth must be high. In addition, storing large volumes of data requires low-cost storage.The servers in a render farm need to access the shared storage concurrently, so the shared storage bandwidth must be high. In addition, storing large volumes of data requires low-cost storage.Inefficient Delivery and Complicated O&M
The construction period is long, the installation and deployment are complicated, and O&M as well as capacity expansion are difficult. Monitoring and alarm management involve large quantity of physical servers. These types of mashups are complicated to manage and consume a lot of manpower.The construction period is long, the installation and deployment are complicated, and O&M as well as capacity expansion are difficult. Monitoring and alarm management involve large quantity of physical servers. These types of mashups are complicated to manage and consume a lot of manpower.
Typical Scenarios
Description
The rendering service is one of the important application fields of high performance computing. The rendering and compute requirements vary with the service scenario. Huawei HPC solution provides end-to-end compute services for enterprises.
Fat nodes of dozens of specifications, such as ECS and BMS compute instances, are supported to meet the various compute to memory ratio requirements of HPC applications.
BMSs provide excellent compute performance as the dedicated physical servers. The service provides the optimal performance without any virtualization loss. Users can use the management console to automatically provision BMSs, thereby meeting HPC service requirements for on-demand elasticity.
Block Storage Service (BSS) is based on the distributed architecture and flexibly scalable. Users can attach EVS disks to an ECS as required to provide storage space for computing and storage nodes. Each disk provides up to 32 TB capacity, 30,000 IOPS, and 1 TB/s throughput.
The solution provides APIs and SDKs for customers to flexibly manage the compute, storage, and network resources on the cloud. Data Express Service (DES) allows customers to import and export data to and from the cloud. Customers can choose the billing model most suitable to them; with pay-per-use convenience so they only pay for the resources they need or they can sign up for a subscription package for even bigger savings.
Solution Architectures
Solution Architecture
Provides high-performance, reliable, and easy-to-use compute, storage, and network services to meet performance requirements in various rendering scenarios, shortening the rendering period and increasing the benefit for the enterprise.
Advantages
Industry-leading performance, providing 100GE InfiniBand compute networks; local 3.2 TB enterprise-level SSD disks, local cache disks with the largest capacity in cloud services; fat node with 96 cores and 2 TB memory
Support for both ECSs and BMSs, meeting various deployment requirements
Based on OpenStack
Comprehensive security protection at seven layers (access, transmission, infrastructure, network, virtualization, data, and management), three-layer Anti-DDoS traffic cleaning, secure network isolation provided by VPC, and access permission control by IAM