Rule-Based Auto-Tagging in OpenSearch: A Flexible Approach To Multi Tenancy
summary
A rule-based auto-tagging system for OpenSearch assigns workload group IDs to incoming requests by evaluating index pattern, username and role and supports exact, pattern and list matching; OpenSearch 3.1 used index-only rules and 3.3 adds username/role. Rules are evaluated all-at-once and the most specific rule wins with a tie‑break hierarchy (username > role > index). A dashboard shows per-group CPU/memory fractions (decimals summing to 100%), completions, rejections and cancellations, and enforcement offers soft vs enforced modes.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:03OpenSearch workload challenges
- 3:56Why workload management
- 7:32Need for auto-tagging
- 9:59Auto-tagging framework
- 16:20Rule matching and precedence
- 19:38Monitoring workload groups
- 21:49Future work and thanks
- 22:33Q&A
related talks
17mFine-Grained Authorization for Secure RAG With OpenSearch
40mUpcoming Changes for the OpenSearch Index Authorization Mechanisms
38mEnabling Secure, Fine-Grained Resource Sharing for Team Collaboration in OpenSearch
13mOpenSearch at Atlassian Scale
21mA Brief Guide to Secure Agentic Knowledge Retrieval
29mThreat Modeling for OpenSearch: Identifying and Mitigating Security Risks
41mSecuring OpenSearch: Hints for avoiding common hassles
25mThe Art of the Possible with OpenSearch Anomaly Detection and UEBA.