Support of heterogeneous hardware
CCMP-EaaS supports the management of complex heterogeneous computing environment and the mixed deployment of generic x86 servers, x86 fat nodes, multi-model GPU servers, ARM servers, FPGA accelerated servers and KNL servers.
Container specialized for HPC
Tailored to the characteristics of high performance computing, ClusterTech developed the specialized container of low performance overhead, flexible management and configurable isolation granularity.
Virtual dynamic cluster
Powered by the self-developed container management technology, the cluster scenarios can be “cloudified”, bringing more flexibility and accessibility of resources to the cluster users.
Rich collection of pre-installed applications
With optimized compilation, a rich collection of research software applications have been pre-installed, thus lowering the user entry barrier. Such out-of-the-box nature enhances the research efficiency.
Segregated operating environment
Applying cutting-edge cloud computing technology, the issue of applications’ dependence on and their conflicts with the operating environment can be avoided by the segregation of the applications from the operating environment.
Various cluster types supported
HPC cluster, big data processing cluster and large-scale training cluster are supported to meet the diverse computing needs of scientific research users.
Supporting both containerization and virtualization
The platform supports both containerization and virtualization technologies at the same time to meet users' needs for ease of use, performance, environment encapsulation, application deployment and other diverse requirements.
Virtual superuser privileges
Normal users can have superuser privileges in a private environment, balancing system security and the flexibility of use.
Multi-dimensional data exchange
In addition to supporting the data interconnect between different application instances under the same user, it also supports collaborative sharing of instance data within the project team, as well as the sharing of application images among users on the platform, making data flow more convenient and the communication of scientific research more smooth.
User management system
For managing the identity authentication, approval, authority control, quota information management, user groups and the relationship with the users.
Container isolation system
Environment isolation is realized through the Namespace technology of Linux core. Resource control is realized through CGroup technology.
Resource scheduling system
According to customer application requirements, the hardware resources such as computing, storage and network are uniformly scheduled and managed. Resource scheduling allocates a complete set of operating environments.
Virtualization management system
The platform supports KVM virtualization technologies and provides more comprehensive operating system support. It also supports full virtual machine lifecycle management and hard disk expansion, snapshot and recovery, etc.
Data management system
Users are entitled to private data directory permissions. It supports the transfer and management of data over a variety of protocols, including HTTP, SFTP, WebDAV, and so on.
Data sharing system
It supports the sharing of private data and shared data sharing within the project team. It also supports data sharing among user local, platform and within the same instance.
GPU management scheduling system
It supports GPU virtualized and containerized scheduling. It also supports 1-N block GPU transmission for a single container and virtual machine, as well as GPU full transmission and multi-machine multi-card GPU training cluster.
Cluster scheduling system
It integrates SLURM, TORQUE and other HPC scheduling systems to achieve the scheduling, distribution and management of cluster tasks. Job submission is carried out through various ways such as web job template, web command line, SSH client, etc.
Big data scheduling system
It integrates various big data processing systems, realizes the processing and analysis of big data tasks, and can directly integrate mature commercial big data platforms.
Image management system
It supports the upload and management of container and virtualized image environments. It also supports users to encapsulate the image environment and to publish and use it through the platform application store.
Report management system
It summarizes and displays through charts user information, platform information, resource status and operational data such as usage statistics of images. All management information is visible here.
Billing management system
It supports separate billing for single users and project teams. It also supports the setting of billing weights according to different resource types. According to the needs of the accounting system customization, it can be customized to interoperate with the financial systems according to the billing requirements.