milvus 2.4.4 cluster read latency affected by insert activity , how to improve?

Question

I notice when inserting 50-60 vectors/sec to certain collection , read latency is increasing significantly only to this collection , reads to other collections in same milvus cluster 2.4.4 are not impacted , I have set no limits for cpu and memory for all nodes in my k8s milvus cluster , but perhaps some internal limits are most probably reached , with the above details could you suggest some suitable internal milvus parameters that may need to be increased so I can reduce read latency to this collection ? Thanks

P.S. I came to think(yet not confirmed) reads somehow wait for inserts to complete , perhaps there is an option to say to search query to not wait for the ongoing inserts ...