Why Is Milvus Search Slow?

How to fix it

Try these in order — the most common fix is first.

1. 1. Locate the stage

   1. Check proxy, queue and execution latency in query node metrics (Milvus bottleneck guide)
   Did this fix it? Yes No

2. 2. Tune parameters

   Copy steps
   1. Tune ef or nprobe against a recall target, not maximum
   2. Split very large NQ requests
   Did this fix it? Yes No

3. 3. Fix data layout

   Copy steps
   1. Create indexes on filtered scalar fields
   2. Check segments are indexed
   3. Allow warm-up after restart; add query nodes
   Did this fix it? Yes No

Symptoms

Rising latency, spikes after restart or scale events, slow filtered queries.

Why it happens

• High ef or nprobe
• Large NQ requests monopolizing query nodes
• Unindexed small segments
• Weak filters or missing scalar indexes
• Cold caches after restart
• Strong consistency waits

This guide also covers

• milvus search slow
• milvus high latency

The same problem in other apps

All slow, laggy or resource-hungry guides →

• Weaviate Out of Memory (OOM Killed) (Weaviate)
• Qdrant Out of Memory (OOMKilled / High RAM Usage) (Qdrant)

Other Milvus errors

• Milvus "collection not loaded" Error
• Milvus Query Node Out of Memory (OOM)
• Milvus "field dimension mismatch" Error
• Milvus Connection Refused / Timeout (19530)
• Milvus etcd and MinIO Dependency Errors
• Milvus "index not exist" Error
• Milvus Search Returns Empty Results After Insert
• How to Cancel Zilliz Cloud (Milvus) Subscription