September 29th, 2024
BlueXP workload factory for generative AI (GenAI) helps customers deploy and manage the infrastructure for retrieval augmented generation (RAG) frameworks. This service now allows you to create point-in-time copies of your RAG data infrastructure at near-instantaneous speed by extending ONTAP Snapshot to Amazon FSx for NetApp ONTAP knowledge bases deployed by workload factory.
Reproducibility of data is critical for developing and deploying production-grade GenAI applications. An error such as ingesting incorrect data into a knowledge base may necessitate reconstructing the entire knowledge base. A developer may want to change the chunking strategy to test how it impacts responses and roll back to a point in time if a change does not yield any benefit. Similarly, knowledge base data must be protected against accidental data loss or malicious data corruption. Before today, you could backup and restore the knowledge base data or rebuild the knowledge base. However these approaches are cumbersome, time consuming and costly. Additionally, there was no way to granularly create a point-in-time copy of a knowledge base as all knowledge bases used a single FSx for ONTAP volume. With knowledge base Snapshot copies, you can now create snapshots of any given knowledge base, automatically or on demand. To roll back, simply select the point-in-time copy and restore the knowledge base to the desired last known good state.
Knowledge base Snapshot copies are now generally available in Workload Factory and can be configured and used via the workload factory API and user interface. Read more in our blog and to get started, sign up with BlueXP workload factory.