|
Work item:
|
F.op-reqts
|
|
Subject/title:
|
Framework and general requirements of observability platform for foundation model training and inference clusters
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2027 (Medium priority)
|
|
Liaison:
|
ISO/IEC JTC1/ SC39
|
|
Supporting members:
|
Zhejiang Lab, China Telecom, China Academy of Information and Communication Technology (CAICT)
|
|
Summary:
|
With the growing demands of foundation model training and inference, the scale and complexity of Artificial Intelligent (AI) computing infrastructure have reached unprecedented levels. To address the increasing operational challenges and improve system reliability, this Recommendation specifies the technical requirements for designing and deploying observability platform in large-scale computing clusters. It establishes a standardized framework to enable consistent monitoring of health, performance and execution behaviors across compute, storage, networking, power and environmental, platform and workloads. It defines requirements for metrics, logs, and traces to support fault detection, root cause analysis, workload optimization, and resilient operations. By promoting interoperable and scalable observability, this Recommendation helps system providers, operators, and vendors build more efficient, reliable, and manageable AI infrastructure.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2025-09-08 15:03:52
|
|
Last update:
2026-09-04 12:10:10
|
