Postgres.AI Chats

Vitaliy Kukharik 1 month ago

Run a series of "over-indexing" tests to measure the potential increase in latency with a large number of indexes.

Details:

Run 20 consecutive select-only pgbench runs (20 indexes), each for 5 minutes.
Before each run, create an index on the pgbench_accounts table (e.g., create index on pgbench_accounts (aid)).
Use a server with 8 vCPUs.
Number of clients as number of CPU cores (-c 8 -j 8)
It's important to avoid CPU saturation, so limit the transactions per second (e.g., 5000 TPS) to keep CPU load below 40%.
Do not reset the cache before each iteration (to eliminate I/O waits).

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

To run a series of "over-indexing" tests in Postgres, we will set up an experiment with the following configuration:

Experiment Configuration

Server Type: n2-standard-8 (8 vCPUs)
Number of Clients: 8 (-c 8 -j 8)
Transaction Limit: 5000 TPS to keep CPU load below 40%
Test Duration: 20 consecutive pgbench runs, each for 5 minutes
Index Creation: Create a new index on the pgbench_accounts table before each run

Steps to Execute

Create a new index on the pgbench_accounts table.
Run the pgbench test for 5 minutes.
Repeat for a total of 20 indexes.

JSON Configuration for the Experiment

Here is the JSON configuration for the experiment:

{
  "MODE": "dedicated",
  "SERVER_TYPE": "n2-standard-8",
  "TEST_RUNS": {
    "workload_pgbench": "-c 8 -j 8 -T 300 -r",
    "pre_shell": "for i in {1..20}; do psql -c 'CREATE INDEX idx_aid_$i ON pgbench_accounts (aid);'; done"
  },
  "VOLUME_SIZE": "100",
  "POSTGRES_VERSION": "16"
}

Confirmation

Would you like me to proceed with running this experiment using the above configuration?

Vitaliy Kukharik 1 month ago | prev | next

No, 20 test runs instead of a cycle and use:

pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

To set up the experiment with the correct configuration for running 20 consecutive pgbench tests, we will use the following details:

Experiment Configuration

Server Type: n2-standard-8 (8 vCPUs)
Number of Clients: 8 (-c 8 -j 8)
Test Command: pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r
Test Duration: 20 consecutive runs, each for 5 minutes
Index Creation: Create a new index on the pgbench_accounts table before each run

JSON Configuration for the Experiment

Here is the updated JSON configuration for the experiment:

{
  "MODE": "dedicated",
  "SERVER_TYPE": "n2-standard-8",
  "TEST_RUNS": {
    "workload_pgbench": "-c 8 -j 8 -S -P 10 -T 300 -R 5000 -r",
    "pre_shell": "for i in {1..20}; do psql -c 'CREATE INDEX idx_aid_$i ON pgbench_accounts (aid);'; done"
  },
  "VOLUME_SIZE": "100",
  "POSTGRES_VERSION": "16"
}

Confirmation

Would you like me to proceed with running this experiment using the above configuration?

Vitaliy Kukharik 1 month ago | prev | next

Example:

{
  "POSTGRES_VERSION": "16",
  "SERVER_TYPE": "n2-standard-8",
  "VOLUME_SIZE": "100",
  "TEST_RUNS": {
    "index_01": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 4 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_02": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 4 -S -P 10 -T 300 -R 5000 -r"
    },
 ...
    "index_20": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 4 -S -P 10 -T 300 -R 5000 -r"
    }
  },
}

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

Here is the updated JSON configuration for running 20 consecutive pgbench tests, each with its own index creation step:

{
  "POSTGRES_VERSION": "16",
  "SERVER_TYPE": "n2-standard-8",
  "VOLUME_SIZE": "100",
  "TEST_RUNS": {
    "index_01": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_02": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_03": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_04": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_05": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_06": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_07": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_08": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_09": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_10": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_11": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_12": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_13": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_14": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_15": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_16": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_17": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_18": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_19": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_20": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    }
  }
}

Next Steps

Would you like me to proceed with running this experiment using the above configuration?

Vitaliy Kukharik 1 month ago | prev | next

yes

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

The experiment has been successfully initiated! You can track the progress and results at the following link: Experiment Pipeline.

Here is the JSON configuration used for the experiment:

{
  "POSTGRES_VERSION": "16",
  "SERVER_TYPE": "n2-standard-8",
  "VOLUME_SIZE": "100",
  "TEST_RUNS": {
    "index_01": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_02": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_03": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_04": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_05": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_06": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_07": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_08": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_09": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_10": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_11": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_12": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_13": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_14": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_15": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_16": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_17": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_18": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_19": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    },
    "index_20": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 5000 -r"
    }
  }
}

If you have any further questions or need assistance, feel free to ask!

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

The experiment has been successfully completed. Here are the results for each pgbench run:

Experiment Results

index_01: 5003.729416 TPS
index_02: 5001.792608 TPS
index_03: 5004.954734 TPS
index_04: 4996.514850 TPS
index_05: 5000.935709 TPS
index_06: 5000.793551 TPS
index_07: 5001.039245 TPS
index_08: 5006.544941 TPS
index_09: 4999.358120 TPS
index_10: 5009.150045 TPS
index_11: 4996.311200 TPS
index_12: 5002.046561 TPS
index_13: 5003.765492 TPS
index_14: 4998.008858 TPS
index_15: 4999.140888 TPS
index_16: 4998.726166 TPS
index_17: 5001.047536 TPS
index_18: 4998.384524 TPS
index_19: 5002.203740 TPS
index_20: 5001.918903 TPS

Analysis

The TPS (Transactions Per Second) values are quite consistent across all runs, indicating that the addition of indexes did not significantly impact the performance in terms of TPS. This suggests that the system was able to handle the additional indexes without a noticeable degradation in performance.

Links to GitLab Jobs

You can review the detailed job logs and results at the following links:

If you have any further questions or need additional analysis, feel free to ask!

Vitaliy Kukharik 1 month ago | prev | next

Visualize the result by latency

[index_01|workload_pgbench] 2024-09-20 16:24:58 latency average = 0.216 ms [index_02|workload_pgbench] 2024-09-20 16:30:19 latency average = 0.221 ms [index_03|workload_pgbench] 2024-09-20 16:35:39 latency average = 0.214 ms [index_04|workload_pgbench] 2024-09-20 16:40:59 latency average = 0.215 ms [index_05|workload_pgbench] 2024-09-20 16:46:20 latency average = 0.216 ms [index_06|workload_pgbench] 2024-09-20 16:51:40 latency average = 0.219 ms [index_07|workload_pgbench] 2024-09-20 16:57:00 latency average = 0.225 ms [index_08|workload_pgbench] 2024-09-20 17:02:21 latency average = 0.226 ms [index_09|workload_pgbench] 2024-09-20 17:07:41 latency average = 0.226 ms [index_10|workload_pgbench] 2024-09-20 17:13:02 latency average = 0.232 ms [index_11|workload_pgbench] 2024-09-20 17:18:22 latency average = 0.236 ms [index_12|workload_pgbench] 2024-09-20 17:23:42 latency average = 0.237 ms [index_13|workload_pgbench] 2024-09-20 17:29:03 latency average = 0.236 ms [index_14|workload_pgbench] 2024-09-20 17:34:23 latency average = 0.243 ms [index_15|workload_pgbench] 2024-09-20 17:39:44 latency average = 0.264 ms [index_16|workload_pgbench] 2024-09-20 17:45:04 latency average = 0.269 ms [index_17|workload_pgbench] 2024-09-20 17:50:24 latency average = 0.272 ms [index_18|workload_pgbench] 2024-09-20 17:55:44 latency average = 0.268 ms [index_19|workload_pgbench] 2024-09-20 18:01:05 latency average = 0.276 ms [index_20|workload_pgbench] 2024-09-20 18:06:25 latency average = 0.283 ms

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

Here is the visualization of the latency results for each pgbench run:

Latency Chart

The chart shows the average latency for each pgbench run, indicating how the latency changes as more indexes are added. If you have any further questions or need additional analysis, feel free to ask!

Vitaliy Kukharik 1 month ago | prev | next

It doesn't seem like there was a significant load on the CPU

06:01:24 PM     CPU     %user     %nice   %system   %iowait    %steal     %idle
06:01:25 PM     all      0.25      0.00      0.13      0.00      0.00     99.62
06:01:26 PM     all     14.46      0.00     11.11      0.00      0.00     74.43
06:01:27 PM     all      8.00      0.00      8.24      0.00      0.00     83.76
06:01:28 PM     all      7.60      0.00      8.93      0.00      0.00     83.47
06:01:29 PM     all      8.14      0.00      8.74      0.00      0.00     83.11
06:01:30 PM     all      8.38      0.00      7.05      0.24      0.00     84.33
06:01:31 PM     all      7.89      0.00      8.37      0.00      0.00     83.74
06:01:32 PM     all      6.98      0.00      9.39      0.00      0.00     83.63
06:01:33 PM     all      8.77      0.00      8.65      0.00      0.00     82.58
06:01:34 PM     all     10.01      0.00      7.75      0.00      0.00     82.24
...
06:06:19 PM     all      9.48      0.00      7.41      0.00      0.00     83.11
06:06:20 PM     all      8.37      0.00      7.76      0.00      0.00     83.87
06:06:21 PM     all      8.69      0.00      8.20      0.00      0.00     83.11
06:06:22 PM     all     11.18      0.00      8.80      0.00      0.00     80.02
06:06:23 PM     all     11.03      0.00      9.19      0.00      0.00     79.78
06:06:24 PM     all      8.14      0.00      7.78      0.00      0.00     84.08
06:06:25 PM     all      7.74      0.00      6.39      0.00      0.00     85.87
Average:        all      8.52      0.00      7.66      0.01      0.00     83.81

let's increase the load 10 times using 50,000 TPS

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

The experiment with increased load (50,000 TPS) has been successfully initiated! You can track the progress and results at the following link: Experiment Pipeline.

Here is the JSON configuration used for the experiment:

{
  "POSTGRES_VERSION": "16",
  "SERVER_TYPE": "n2-standard-8",
  "VOLUME_SIZE": "100",
  "TEST_RUNS": {
    "index_01": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_02": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_03": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_04": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_05": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_06": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_07": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_08": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_09": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_10": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_11": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_12": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_13": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_14": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_15": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_16": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_17": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_18": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_19": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    },
    "index_20": {
      "pre_sql": "create index on pgbench_accounts (aid);",
      "workload_pgbench": "pgbench -c 8 -j 8 -S -P 10 -T 300 -R 50000 -r"
    }
  }
}

If you have any further questions or need assistance, feel free to ask!

Vitaliy Kukharik 1 month ago | prev | next

result?

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

Experiment Results with Increased Load (50,000 TPS)

Here are the results for each pgbench run with the increased load:

index_01: 49988.113066 TPS
index_02: 49990.321320 TPS
index_03: 50020.595981 TPS
index_04: 50015.570509 TPS
index_05: 49981.587002 TPS
index_06: 49997.997562 TPS
index_07: 49982.745659 TPS
index_08: 49971.338029 TPS
index_09: 50000.173694 TPS
index_10: 50004.386136 TPS
index_11: 50025.779947 TPS
index_12: 49994.024354 TPS
index_13: 49997.509827 TPS
index_14: 49996.337621 TPS
index_15: 49215.109826 TPS
index_16: 44961.902198 TPS
index_17: 40764.334340 TPS
index_18: 39368.920880 TPS
index_19: 38428.076467 TPS
index_20: 37935.121960 TPS

Analysis

The TPS values are initially consistent but start to drop significantly after the 14th index. This indicates that the system begins to experience performance degradation as more indexes are added, especially under the increased load of 50,000 TPS.

Next Steps

Would you like to visualize these results or perform any further analysis?

Vitaliy Kukharik 1 month ago | prev | next

here is the result of the latency, visualize it:

[index_01|workload_pgbench] 2024-09-20 18:40:00 latency average = 0.456 ms
[index_02|workload_pgbench] 2024-09-20 18:45:20 latency average = 0.496 ms
[index_03|workload_pgbench] 2024-09-20 18:50:40 latency average = 0.575 ms
[index_04|workload_pgbench] 2024-09-20 18:56:01 latency average = 0.669 ms
[index_05|workload_pgbench] 2024-09-20 19:01:21 latency average = 0.834 ms
[index_06|workload_pgbench] 2024-09-20 19:06:42 latency average = 0.847 ms
[index_07|workload_pgbench] 2024-09-20 19:12:02 latency average = 1.076 ms
[index_08|workload_pgbench] 2024-09-20 19:17:23 latency average = 11.931 ms
[index_09|workload_pgbench] 2024-09-20 19:22:43 latency average = 1.480 ms
[index_10|workload_pgbench] 2024-09-20 19:28:03 latency average = 1.944 ms
[index_11|workload_pgbench] 2024-09-20 19:33:24 latency average = 2.794 ms
[index_12|workload_pgbench] 2024-09-20 19:38:44 latency average = 3.843 ms
[index_13|workload_pgbench] 2024-09-20 19:44:04 latency average = 9.203 ms
[index_14|workload_pgbench] 2024-09-20 19:49:24 latency average = 18.591 ms
[index_15|workload_pgbench] 2024-09-20 19:54:45 latency average = 2444.281 ms
[index_16|workload_pgbench] 2024-09-20 20:00:05 latency average = 15270.280 ms
[index_17|workload_pgbench] 2024-09-20 20:05:25 latency average = 27859.220 ms
[index_18|workload_pgbench] 2024-09-20 20:10:45 latency average = 31619.704 ms
[index_19|workload_pgbench] 2024-09-20 20:16:05 latency average = 35224.589 ms
[index_20|workload_pgbench] 2024-09-20 20:21:25 latency average = 36169.644 ms

Postgres.AI 1 month ago | prev | next | oai/gpt-4o

Latency Visualization

Here is the visualization of the latency results for each pgbench run with the increased load:

Latency Chart

The chart shows the average latency for each pgbench run, indicating how the latency changes as more indexes are added under the increased load of 50,000 TPS. If you have any further questions or need additional analysis, feel free to ask!