Skip to content

ECO-AIM-AI-015 — No batching of vector search queries

  • Category: AI/ML (AIM)
  • Family: AI (AI)
  • Layer: AI
  • Tier: 2
  • Severity: note
  • Tags: ai, vector-search
  • Legacy ID: ECO-AI-015

Summary

Unbatched vector queries increase overhead and reduce throughput.

Rationale

Batching reduces per-request overhead for retrieval.

Impact

{
  "confidence": 0.55,
  "notes": "Depends on query load.",
  "type": "cpu"
}

Detection

{
  "languages": [
    "infra"
  ],
  "method": "trace"
}

Remediation

{
  "guidance": "Batch retrieval queries where possible.",
  "tradeoffs": "May add slight latency."
}

Ontology

{
  "system_layers": [
    "ai"
  ]
}