Simplify Multi-Step, Long-Running, Stateful Application Orchestration
CloudFlow is a fully managed serverless cloud service used to coordinate the execution of multiple distributed tasks. It is committed to simplifying tedious tasks, such as task coordination, state management, and error handling required to develop and run business processes. It enables you to focus on business logic development.
You can arrange distributed tasks in sequence, branch, and parallel. The service will reliably coordinate task execution according to the set sequence and track the state transition of each task. Then, it will execute custom retry logic (whenever necessary) to ensure the smooth completion of the workflow.
The Distributed Component
Orchestrate applications written in different infrastructures, networks, and languages, and eliminate the transition from a hybrid cloud and proprietary cloud to a public cloud or evolve from a single architecture to a microservice architecture.
Reduce the Amount of Code
It helps you manage the logic of your application and supports multiple control primitives, such as sequence, branch, and parallel, to minimize repeated orchestration code in your business.
Improve Fault Tolerance
Manage process status, set checkpoints, and process reboots to ensure that your applications are performing step-by-step as expected. Built-in try/catch, retry, and rollback functions can automatically handle exception errors.
Features
Operation & Maintenance Fully-Managed and Pay-As-You-Go
Frees operation and maintenance personnel from infrastructure maintenance and provides secure, highly available, highly fault-tolerant, and highly durable services. You only need to pay the step conversion fee; no use means no additional fees!
Service Orchestration Capabilities
It helps you separate process logic from task execution, supports multiple control primitives, such as sequence, branch, and parallel. It also has built-in error retry and capture capabilities, saving time for writing orchestration code.
Support Long Running Process
Track the whole process and ensure the completion of process execution for a long time. Some processes may need to be executed for hours, days, or months, such as operation and maintenance, related pipeline, and mail promotion processes.
The Process State Management
Manage all states in the execution of a process, including tracking the execution steps where it is located, and the input and output stored between steps. You do not need to manage the process state yourself or build complex state management into tasks.
Visual Monitoring
Provides a visual interface to define workflows and view execution status. Status includes input and output. It helps you to quickly identify the fault location and quickly eliminate the fault problem.
The Distributed Component
Coordinates applications written in different languages on different infrastructures and within different networks. Whether the application is smoothly transitioning from private cloud/Private Cloud to hybrid cloud or public cloud, or evolving from a single architecture to a microservice architecture, CloudFlow can play a coordinating role.
Pricing
You only need to pay according to the usage. Pay according to the number of steps in the workflow execution and charge the state conversion fee of function computing, ECS, container, and other resources in the workflow. Then, charge by time. The end of execution shall not charge fees any longer if the first 5,000 times of the main account per month are free, but will charge 0.025 USD/thousand times afterward.
Serverless service automatic expansion saves you from managing hardware budgets and expansion.
How do I use CloudFlow?
1
Step 1: Create a Workflow Definition
Use definitions to describe the order or logic of execution between tasks.
2
Step 2: Prepare Workflow Tasks
You need to implement each task Worker defined in the workflow. CloudFlow supports several different types of workers.
3
Step 3: Perform Workflow Tasks
You can call CloudFlow, Execute Workflow, and APIs to execute workflow applications.
4
Step 4: Track, Debug, and Monitor Running Status
Use the visual monitoring provided by CloudFlow to observe the running status of tasks.
Scenarios
Use multi-task orchestration to reduce process failure rate and improve multimedia throughput
Using CloudFlow can help you organize multiple tasks, such as transcoding, frame interception, face recognition, speech recognition, and audit upload, into a complete processing flow. Submit a media processing (IMM) task (or self-built processor) through function compute to produce an output that meets business requirements. The execution error of each task can be reliably retried, greatly improving the processing throughput of multimedia tasks.
Able to Resolve
Multi-Task Orchestration
Layout the process steps, such as transcoding, frame interception, face recognition, speech recognition, and audit upload.
Solving Process Failure
The system will retry the execution error of the task multiple times.
Simplify large-scale computing orchestration, optimize costs, and reduce total process duration
CloudFlow can orchestrate multiple batch computing distributed jobs in series or in parallel to reliably support large-scale computing with long execution time and large concurrency. For example, in the analysis of gene data, align the gene sequences, conduct mutation analysis on all chromosomes in parallel, and finally aggregate the chromosome data to produce results. CloudFlow submits batch computing jobs of different CPU, memory, and bandwidth specifications based on specified dependencies to improve execution reliability, improve resource utilization, and optimize costs.
Able to Resolve
Simplify the Large-scale Computing Orchestration
CloudFlow orchestrates multiple batch computing distributed jobs in series or in parallel and supports computing with long execution time and large concurrency.
Shorten the Total Process Duration
Submit batch computing jobs of different specifications based on task dependencies to improve the reliability of parallel execution.
Reliable scheduling of massive computing resources and personalized orchestration of big data processing tasks
Using CloudFlow can quickly build a highly available data processing pipeline. For example, metering data from different data sources is collected to log service and the timer of function computing triggers the workflow every hour. CloudFlow uses function computing to perform parallel processing on metering data of multiple shards, and the results are written back to log service, respectively. Then, all Shard generated files can be aggregated, written to table store, and a bill is generated for each user. CloudFlow supports retrying a single step in the process to reduce the probability of process failure. CloudFlow also supports dynamic parallel task execution and high scalability of data processing capability.
Able to Resolve
Resolve Concurrency Implementation Issues
Supports dynamic parallel task execution and high scalability of data processing capabilities.
Visual interface for improving O&M efficiency and reliability and process progress
The common challenges of automated O&M include the complexity of steps, the low reliability of stand-alone scripts, the complexity of dependencies, and the progress cannot be visualized. The CloudFlow + function compute process can address these challenges. For example, automated software deployment starts by building a Docker container, uploading a container image, and tracking each node to pull down the image. Start the container of the new image. Then, logs generated by each function step are saved to the log service for query and sharing. Compared with single-machine O&M scripts, workflow-based automation tools have the features of high availability, built-in error handling mechanism, and graphical progress.
Able to Resolve
Resolve Process Progress Visualization Issues
Automated O&M procedures take a variety of time, and the reliability of stand-alone scripts is low. Common challenges, such as complex dependencies, cannot be visualized.
Related Products
You can use CloudFlow to coordinate the execution of multiple distributed tasks.
It can be used to integrate computing services, such as function computing and batch computing, and to enable tasks running on ECS and container services to act as executors through Poll. It provides external event triggering capabilities and integrated message service.
Function Compute
Pay-As-You-Go Serverless Computing Platform
ECS
Virtual Cloud server with fast memory and latest Intel CPU
Message Service
Help application developers freely transfer information and send notifications between application components
Upgraded Support For You
1 on 1 Presale Consultation, 24/7 Technical Support, Faster Response, and More Free Tickets.
1 on 1 Presale Consultation
24/7 Technical Support
6 Free Tickets per Quarter
