Near-Offline Resource Hosting Model
Last updated: 2025-06-17 16:36:30
Overview
The Near Offline Resource Hosting Model is a low-cost serverless computing solution provided by Serverless Cloud Function (SCF). It intelligently schedules a pool of bidding instance resources across global multiple regions, significantly reducing computing costs while maintaining the full features of serverless. This mode supports functions to run in the near offline cluster on the platform, and with the aid of a dynamic scheduling algorithm, automatically assigns requests to the optimal resource nodes, achieving cost-effective computing services.
Notes:
The resource pool has a risk of passive release. It is recommended for near-offline task scenes with high fault tolerance. A retry mechanism needs to be considered to ensure business continuity.
The concurrency limit of the function can be adjusted by contacting the function side to enhance the overall concurrency capability.
Application Scenarios
The new model launched by the function platform is suitable for nearline scenarios (such as Kafka task queues) and offline scenarios (such as dial testing and stress testing). Tasks are submitted through the function platform's trigger. The platform manages the queue management, resource supply, and result reporting of task execution throughout the process. It ensures high availability and achieves an overall cost lower than that of bidding instances.
Nearline Scenarios
Generally, data is processed quasi-real-time based on the message queue. It can respond in seconds or minute-level, without requiring immediate returned results, for example, media transcoding, data processing, model inference, message notifications.
Offline Scenarios
High computing throughput, bulk initiation, no requirement for immediate processing, with a certain flexible expectation for processing time. High resource usage rate during operation. For example, stress tests/dial tests, data archiving, automation (RPA), other scheduled tasks.
Billing Mode
When the namespace type is "Nearline Resource Pool Type", it supports pay-as-you-go (postpaid) . The original resource usage fee under this namespace will be split into VCPU usage fee and memory usage fee. Other billing items and the hosting of the default resource pool remain unchanged.
Product Pricing
The cost of SCF consists of the following components. Pricing is as follows:
VCPU usage fee: 0.00000124 USD/VCPU*second (0.0124 USD/10,000 VCPU*seconds)
Usage fees for memory: 0.00000012 USD/GB*s (0.0012 USD/10,000 GB*s)
API call fee: 0.002 USD per 10,000 calls.
public network outbound traffic fee: Different pricing in each region. In the Chinese mainland, it is 0.12 USD/GB.
Idle provisioned concurrency fee: 0.00000847 USD/GBs (0.0847 USD/10,000 GBs). Please refer to billing detail and billing example.
Web functions and event functions share the same pricing. For web functions using the default trigger, additional response traffic for web functions will be incurred. For details on web function billing, see Web Function Billing Description.
Features and Advantages
Compared with the default resource hosting mode, the Near Offline Resource Hosting Model has following strengths:
Extremely low cost expenditure: By leveraging the bidding instance resource pool across global multiple regions, computing costs can be reduced by over 90%. It supports the time-sharing strategy, automatically utilizes the tidal characteristics of resources during idle periods such as nighttime, retains the pay-as-you-go mode, and avoids resource waste due to idleness.
Complete Serverless experience: Zero modification for business development; APIs and development methods fully compatible with the default resource pool. Retained maintenance-free features: Basic capabilities such as automatic monitoring, logs, and alarms are fully inherited. Intelligent auto scaling: Supports scaling out in seconds and can handle tens of thousands of concurrent requests in a single region.
Intelligent resource scheduling: Dynamic routing algorithm: Automatically select the best resource node (based on cost/delay/stability); Hybrid deployment support: Can be used in conjunction with the standard resource pool in proportion; Automatic compensation for interruptions: Built-in dual guarantees mechanism of request retry and dead letter queue.
Operation Steps
Create and Bind a Function Namespace to Nearline/Offline Resource Pools
1. Log in to Serverless Console, click Function Service in the left sidebar.
2. On the Function Service page, select the expected region for function creation at the top, click the ⚙️ icon on the right of the namespace, and enter namespace management. As shown below:
3. In the namespace pop-up window, click add namespace to enter the namespace creation page. As shown below:
4. In the Resource Hosting Mode options, select Near Offline and complete the creation.
5. Switch to the created function namespace to create a function and start using it.
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No
Feedback