What Is Workflow?

MLOps Overview

Understanding MLOps is essential before learning about workflows.

Machine Learning Operations (MLOps) are a set of practices with machine learning (ML) and DevOps combined. The ML development process consists of project design, data engineering, model building, and model deployment. AI development is not a unidirectional pipeline job. During development, multiple iterations of experiments are performed based on the data and model results. To achieve better model results, algorithm engineers perform diverse data processing and model optimization based on the data features and labels of existing datasets. Traditional AI development ends with a one-off delivery of the final model output by iterative experimentation. As time passes after an application is released however, model drift occurs, leading to worsening effects when applying new data and features to the existing model. Iterative experimentation of MLOps forms a fixed pipeline which contains data engineering, model algorithms, and training configurations. You can use the pipeline to continuously perform iterative training on data that is being continuously generated. This ensures that the AI application of the model, built using the pipeline, is always in an optimum state.

Figure 1 MLOps

An entire MLOps link, which covers everything from algorithm development to service delivery and O&M, requires an implementation tool. Originally, the development and delivery processes were conducted separately. The models developed by algorithm engineers were delivered to downstream system engineers. In this process, algorithm engineers are highly involved, which is different from MLOps. There are general delivery cooperation rules in each enterprise. When it comes to project management, working process management needs to be added to AI projects as the system does not simply build and manage pipelines, but acts as a job management system.

The tool for the MLOps link must support the following features:

Process analysis: Accumulated industry sample pipelines help you quickly design AI projects and processes.
Process definition and redefinition: You can use pipelines to quickly define AI projects and design workflows for model training and release for inference.
Resource allocation: You can use account management to allocate resource quotas and permissions to participants (including developers and O&M personnel) in the pipeline and view resource usage.
Task arrangement: Sub-tasks can be arranged based on sub-pipelines. Additionally, notifications can be enabled for efficient management and collaboration.
Process quality and efficiency evaluation: Pipeline execution views are provided, and checkpoints for different phases such as data evaluation, model evaluation, and performance evaluation are added so that AI project managers can easily view the quality and efficiency of the pipeline execution.
Process optimization: In each iteration of the pipeline, you can customize core metrics and obtain affected data and causes. In this way, you can quickly determine the next iteration based on these metrics.

Workflow Overview

A workflow is a pipeline tool developed based on service scenarios for deploying models or applications. In ML, a pipeline may involve data labeling, data processing, model development and training, model evaluation, application development, and application evaluation.

Figure 2 Workflow

Different from traditional ML-based model building, workflows can be used to develop production pipelines. Based on MLOps, workflows enable runtime recording, monitoring, and continuous running. The development and continuous iteration of a workflow are separated in products based on roles and concepts.

A pipeline consists of multiple phases. The functions required by the pipeline and the function parameters are called through workflow SDKs. When developing a pipeline, you can use SDKs to describe phases and the relationships between phases. Developing a pipeline is the development state of the workflow. After a pipeline is determined, you can consolidate and provide it for others to use. You do not need to pay attention to what algorithms are used in the pipeline or how the pipeline is implemented. Instead, you only need to check whether the models or applications produced by the pipeline meet the release requirements. If not, you need to check whether the data and parameters need to be adjusted for iteration. Using such a consolidated pipeline is the running state of the workflow.

The development and running states of a workflow are as follows:

Development state: Workflow Python SDKs are used to develop and debug a pipeline.
Running state: You can configure and run a produced pipeline in visualized mode.

Leveraging DevOps principles and practices, workflows orchestrate ModelArts capabilities to help you efficiently train, develop, and deploy AI models.

Different functions are implemented in the development and running states of a workflow.

Workflow Development State

Based on service requirements, you can use Python SDKs provided by ModelArts workflows to offer each ModelArts capability as a step in a pipeline. This is a familiar and flexible development mode for AI developers. Python SDKs support:

Development and building: You can use Python code to create and orchestrate workflows with flexibility.
Commissioning: The debug and run modes are supported. The run mode supports partial execution and fully execution of a workflow.
Publishing: The debugged workflows can be fixed and published to the running state for configuration and execution.
Experiment record: for persistence and the management of experiments.
Sharing: Workflows can be published to AI Gallery as assets and shared with other users.

Workflow Running State

Workflows are executed in visualized mode. You only need to pay attention to some simple parameter settings to start a workflow. Running workflows are released from the development state or subscribed to from AI Gallery.

Running workflows are released from the development state or subscribed to from AI Gallery.

A running workflow supports:

Unified configuration management: The parameters and resources required for a workflow are centrally managed.
Easy-to-use operations: You can start, stop, retry, copy, and delete workflows.
Running record: records historical running parameters and statuses of the workflow.

Workflow Components

A workflow is the description of a directed acyclic graph (DAG). You can develop a DAG through a workflow. A DAG consists of phases and the relationships between phases. To define a DAG, specify the execution content and sequence on phases. A green rectangle indicates a phase, and the link between phases shows the phase relationship. A DAG is actually an ordered job execution template.