Application Scenarios

Big Data and AI Computing

Currently, most big data and AI training applications (such as TensorFlow and Caffe) run in containerized mode. These applications are GPU intensive and require high-performance network and storage. In addition, as these applications are task-based, resources must be quickly allocated upon task creation and released upon task completion.

The following CCI features make it suitable for running these applications:

Accelerated computing with heterogeneous GPUs and Ascend chips (in-house AI chips)
Large-scale, high-concurrency container creation and management
On-demand usage and billing

Figure 1 Big data and AI computing
Click to enlarge

Scientific Computing

Scientific R&D in fields such as genomics and drug development requires high-performance and high-density computing. In addition, scientific computing is generally task-based and resources need to be quickly allocated and released. Therefore, a low-cost computing platform with automated O&M is required.

The following CCI features make it suitable for computing in this scenario:

High-performance computing and network, and high I/O storage
Resource scaling in seconds minimizes resource consumption
No O&M required for clusters and servers, greatly reducing O&M costs
On-demand usage and billing

DevOps/Continuous Delivery

Software development enterprises need a complete DevOps process from code submission to application deployment to improve the development efficiency. DevOps processes such as continuous integration/continuous delivery (CI/CD) are generally task-based computing and require quick resource allocation and release.

The following CCI features make it suitable for computing in this scenario:

Automation for the entire CI/CD process, with no cluster creation and maintenance required
Image-based delivery, allowing for consistency between the development and production environments
On-demand usage and billing

Figure 2 DevOps/Continuous delivery

Services with Fluctuating Traffic

Some types of applications, such as live video, media information, e-commerce, and online education, have obvious service peaks and troughs. For these applications, resources need to be expanded rapidly during peak hours without breaking the bank.

The following CCI features make it suitable for these applications:

Fast scaling: CCI can quickly take over services from CCE to ensure uptime during peak hours.
Low-cost, flexible billing modes: When services are stable, they can be run on CCE and be periodically billed. During peak hours, additional services can be run on CCI and be billed based on the usage. This mode greatly reduces costs.

Figure 3 Auto scaling