|
Work item:
|
Y.MLPaaS-reqts
|
|
Subject/title:
|
Cloud Computing - Functional requirements of PaaS for machine learning applications
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2028-Q2 (No priority specified)
|
|
Liaison:
|
ITU-T All Study Groups, ITU-R SG6, ISO/IEC JTC 1/SC 27, SC 38, SC 42
|
|
Supporting members:
|
Kyung Hee University, ETRI, Korea (Rep. Of), China Telecom
|
|
Summary:
|
The growing demand for machine learning (ML) applications is leading cloud service providers (CSPs) to integrate such applications into their cloud platforms. Software solutions such as ML learning model optimizations and workload scheduling, along with hardware solutions such as GPUs, TPUs and NPUs help enhance the performance and efficiency in development and deployment of ML applications. However, traditional cloud service models such as Machine Learning as a Service (MLaaS) or generic cloud-native PaaS platforms lack fine-grained support for the full life-cycle of ML application development and deployment. This Recommendation focuses on specifying the functional requirements of a dedicated PaaS for ML applications (MLPaaS), which supports both development and deployment workflows along with ML models tailored for ML applications. MLPaaS also takes into account challenges related to application-specific requirements, heterogeneous infrastructures, and the complexity of orchestrating ML workflows. Therefore, we propose developing a draft recommendation titled functional requirements of PaaS for ML applications.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2025-07-30 13:36:16
|
|
Last update:
2025-07-30 13:37:47
|
