We've increased data disk requirements for the multi-node HA-ready production profile. The requirements is now 2 TiB SSD. Note that these new requirements must be met by all 2021.10 versions.
Wondering why we did that?
Data disk is the storage for the cluster runtime artifacts, such as containers, logs etc. To ensure availability and IO throughput, the data is replicated across the nodes, hence the 2TB required per node.
This is dictated by the size of the AI models, training data, and exhaust generated by AI Center.
- Fixed a security issue regarding
- Fixed a vulnerability due to SpringShell.
- Fixed an issue where a pipeline could not be stopped (killed) properly.
- Fixed a bug that caused Data Labeling sessions to not be created with Orchestrator version 21.4 or lower.
Click to learn more...
This version of AI Center is available in different deployment models:
- standalone AI Center
- AI Center service which is part of Automation Suite.
The product is similar enough across deployment types to share the same documentation.
But differences do exist. When certain information applies to only one of the deployments, we use the following labels:
- - only applies for standalone AI Center and does not apply for Automation Suite AI Center.
- - only applies for Automation Suite AI Center and does not apply for standalone AI Center.
Whenever there is no label, the information applies for both deployment types.
Updated 8 months ago