All Products
Search
Document Center

Platform For AI:Data security

Last Updated:Oct 31, 2024

Data transmission security

To enhance data transmission security for a model service that is deployed in Platform for AI (PAI), you can access the service over a virtual private cloud (VPC) by using the official SDK for Python or implementing custom logic. VPCs are isolated from each other, which ensures private high-speed access.

For more information about VPC security, see Security system overview.

Data storage security

PAI provides a dataset management module that allows you to create datasets by using public datasets or data that is stored in Alibaba Cloud storage services. The module also allows you to scan Object Storage Service (OSS) folders to generate index datasets that you can use for intelligent labeling and model training.

To ensure the storage security of data files such as datasets and models, we recommend that you use the following Alibaba Cloud storage services:

Data storage service

Related topic

OSS

Security and compliance

File Storage NAS

Functions and features

Cloud disks

Disks

MaxCompute

FAQ about security configurations

Data backup and restoration

PAI allows you to use images to back up and restore user environments and datasets. For example, you can create an image to save the data, engineering environments, and configurations of an instance. This way, you can deploy the image based on your business requirements, which provides a safeguard against data loss. When you use Alibaba Cloud storage services to store data or datasets, backup and restoration is guaranteed by the corresponding service.

Data storage service

Related topic

OSS

Data backup and recovery

File Storage NAS

Back up and restore files

MaxCompute

Backup and restoration

When you use PyTorch to train a foundation model, you can use EasyCkpt to save and resume the model training progress to minimize data loss and waste of resources. EasyCkpt is a high-performance checkpoint framework provided by PAI and is compatible with Megatron and DeepSpeed, which are popular model training frameworks. For more information, see Use EasyCkpt to save and resume foundation model trainings.