Scaling qps

Skill qdrant/skills/skills/qdrant-scaling/scaling-qps

Guides Qdrant query throughput (QPS) scaling. Use when someone asks 'how to increase QPS', 'need more throughput', 'queries per second too low', 'batch search', 'read replicas', or 'how to handle more concurrent queries'.From its SKILL.md

Install

npx -y skills add qdrant/skills --skill scaling-qps

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

3.6 KB, as published. Nobody here has run it

Scaling for Query Throughput (QPS)

Throughput scaling means handling more parallel queries per second. This is different from latency - throughput and latency are opposite tuning directions and cannot be optimized simultaneously on the same node.

High throughput favors fewer, larger segments so each query touches less overhead.

Performance Tuning for Higher RPS

Use fewer, larger segments (default_segment_number: 2) Maximizing throughput
Enable quantization pinned in RAM to reduce disk IO: memory: pinned on Qdrant 1.19 or newer, always_ram: true on 1.18 or older Quantization
Use batch search API to amortize overhead Batch search

Minimize impact of Update Workloads

Configure update throughput control (v1.17+) to prevent unoptimized searches degrading reads Low latency search
Set optimizer_cpu_budget to limit indexing CPUs (e.g. 2 on an 8-CPU node reserves 6 for queries)
Configure delayed read fan-out (v1.17+) for tail latency Delayed fan-outs

Horizontal Scaling for Throughput

If a single node is saturated on CPU after applying the tuning above, scale horizontally with read replicas.

Shard replicas serve queries from replicated shards, distributing read load across nodes
Each replica adds independent query capacity without re-sharding
Use replication_factor: 2+ and route reads to replicas Distributed deployment

See also Horizontal Scaling for general horizontal scaling guidance.

Disk I/O Bottlenecks

If it is not possible to keep all vectors in RAM, disk I/O can become the bottleneck for throughput. In this case:

Upgrade to provisioned IOPS or local NVMe first. See impact of disk performance to vector search in Disk performance article
Use io_uring on Linux (kernel 5.11+) io_uring article
In case of quantized vectors, prefer global rescoring over per-segment rescoring to reduce disk reads. Example in the tutorial
Configure higher number of search threads to parallelize disk reads. Default is cpu_count - 1, which is optimal for RAM-based search but may be too low for disk-based search. See configuration reference
If still saturated, scale out horizontally (each node adds independent IOPS)

What NOT to Do

Do not expect to optimize throughput and latency simultaneously on the same node
Do not use many small segments for throughput workloads (increases per-query overhead)
Do not scale horizontally when IOPS-bound without also upgrading disk tier
Do not run at >90% RAM (OS cache eviction = severe performance degradation)

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most rag retrieval skills give

Counted across 199 of the 213 authors here whose files we hold, read 2026-09-06

Enable caching for frequent queriesin 14 of 199, across 5 files
Enable MMR for diverse resultsin 12 of 199, across 5 files
Enable binary quantization to reduce memoryin 11 of 199, across 4 files
Initialize the database with dimensions matching the embedding modelin 11 of 199, across 4 files
Start the similarity threshold at 0.7in 11 of 199, across 4 files
Check database statistics when diagnosing slow searchin 11 of 199, across 4 files
Export and import vectors as JSONin 10 of 199, across 3 files
Match index dimension to the embedding modelin 10 of 199, across 9 files
Order filters cheap before expensivein 9 of 199, across 2 files
Generate a runnable scaffold in the user's stackin 9 of 199, across 2 files
Recommend multi-action scoring when frequent tuning is expectedin 9 of 199, across 2 files
Batch store documents for bulk insertsin 9 of 199, across 2 files

Said here and by no other author read

Limit background optimization with optimizer_cpu_budget
Use fewer, larger segments for throughput
Enable quantization pinned in RAM
Use the batch search API to amortize overhead
Configure update throughput control to protect reads
Configure delayed read fan-out for tail latency

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.