Function Delivery Network: A Spatial-Temporal Execution Orchestrator for Optimizing Serverless Computing
We present Function Delivery Network (FDN), a novel spatial-temporal execution orchestrator designed to address key limitations in current serverless computing models. The FDN introduces several innovations, including a multi-tenant serverless model, a secure and reusable functional context, and dis...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11053862/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | We present Function Delivery Network (FDN), a novel spatial-temporal execution orchestrator designed to address key limitations in current serverless computing models. The FDN introduces several innovations, including a multi-tenant serverless model, a secure and reusable functional context, and distributed shared memory, to optimize resource allocation and improve performance in high-concurrency and globally distributed scenarios. We implement the FDN on a major cloud platform and evaluate its performance against traditional Function-as-a-Service (FaaS) execution using a variety of benchmark functions. Our results demonstrate significant improvements in resource utilization, execution time, and request completion rates. The FDN reduces function instance allocation by up to 97.82%, improves median response times by 45.45%, and maintains higher request completion rates at high concurrency levels compared to native FaaS execution. The FDN’s adaptive execution window mechanism allows for fine-tuned optimization based on function characteristics and workload patterns. This approach effectively addresses challenges such as cold starts, inefficient resource allocation, and scalability limitations in current serverless platforms. By providing a more efficient and scalable model for serverless computing, the FDN enables more cost-effective and performant cloud-native applications, particularly in scenarios involving high concurrency and global distribution. |
|---|---|
| ISSN: | 2169-3536 |