tencent cloud

Query Performance Test
Download PDF
Last updated:2026-01-23 17:32:06
Query Performance Test
Last updated: 2026-01-23 17:32:06
Download PDF

Test Objective

This document aims to evaluate the query throughput (QPS) and response latency performance of the database by adjusting concurrent pressure.

Performance Metrics

Metric Name
Description
Calculation Formula/Unit
QPS
Number of completed queries per second
Total queries/Total time (s)
Average Latency
Average response time of all query requests
Total time (ms)/Total queries
Minimum Latency
Shortest response time among all query requests
Millisecond (ms)
Maximum Latency
Longest response time among all query requests
Millisecond (ms)

Test Query Types

Query Type
Query Scope
single-groupby-1-1-1
For a single metric on one host, perform a MAX aggregation every 5 minutes over a period of 1 hour.
single-groupby-1-8-1
For a single metric on 8 hosts, perform a MAX aggregation every 5 minutes over a period of 1 hour.
single-groupby-5-1-1
For one host, collect 5 metrics and perform a MAX aggregation every 5 minutes over a period of 1 hour.
cpu-max-all-1
Perform a MAX aggregation on all CPU metrics of a single host within 1 hour.

Test Procedure

1. Use the tsbs_generate_queries command to specify the query type and generate query packages. The following example uses the single-groupby-1-1-1 query type:
 ./tsbs_generate_queries --use-case="cpu-only" --seed=123 --scale=10000 \\
    --timestamp-start="2025-01-01T00:00:00Z" \\
    --timestamp-end="2025-01-01T1:00:01Z" \\  # Query data for a 1-hour period.
    --queries=100000 \\               # Generate 100,000 queries.
    --query-type="single-groupby-1-1-1" \\  # The query type.
    --format="influx" \\
    | gzip > query/influx-100000queries-single-groupby-1-1-1-12h.gz
2. Based on the query request (influx-100000queries-single-groupby-1-1-1-12h.gz) generated in the previous step, use tsbs_run_queries_influx to query the time series database. The following example uses a concurrency of 320:
cat query/influx-100000queries-single-groupby-1-1-1-12h.gz | gunzip | ./tsbs_run_queries_influx \\
    --urls=http://XX.XX.XX.XX:8086 \\   # The CTSDB 3.0 cluster address.
    --workers=320 \\                  # Set the number of concurrent workers to 320.
    --db-name=tsbs_test \\            # The database to query.
    --print-interval 10000           # Print the progress every 10,000 queries.

Test Results

Query Performance of single-groupby-1-1-1

For a single metric on one host, perform a MAX aggregation every 5 minutes over a period of 1 hour.

Test Concurrency
QPS
Minimum Latency (ms)
Average Latency (ms)
Maximum Latency (ms)
80
9413.97
6.56
8.48
61.98
160
16417.66
6.62
9.70
58.04
240
19065.37
7.71
12.52
84.77
320
21080.60
7.08
15.09
82.44

Query Performance of single-groupby-1-8-1

For a single metric on 8 hosts, perform a MAX aggregation every 5 minutes over a period of 1 hour.

Test Concurrency
QPS
Minimum Latency (ms)
Average Latency (ms)
Maximum Latency (ms)
80
5743.46
7.71
13.86
41.36
160
6466.57
9.58
24.58
101.25
240
6790.33
10.09
35.11
97.95
320
6853.73
11.96
46.37
140.94

Query Performance of single-groupby-5-1-1

For one host, collect 5 metrics and perform a MAX aggregation every 5 minutes over a period of 1 hour.

Test Concurrency
QPS
Minimum Latency (ms)
Average Latency (ms)
Maximum Latency (ms)
80
7773.55
7.47
10.21
63.98
160
12877.07
8.34
12.29
47.08
240
14664.10
8.89
16.15
77.00
320
15198.80
8.91
20.77
71.86

Query Performance of cpu-max-all-1

Perform a MAX aggregation on all CPU metrics of a single host within 1 hour.

Test Concurrency
QPS
Minimum Latency (ms)
Average Latency (ms)
Maximum Latency (ms)
80
7323.38
8.31
10.86
39.25
160
12197.97
9.48
12.98
56.18
240
14600.77
10.11
16.23
105.48
320
15641.33
10.62
20.16
99.38
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No

Feedback