{Name: "subquery_having_2", Age: 20},
		{Name: "subquery_having_3", Age: 30},
		{Name: "subquery_having_4", Age: 40},
	}
	DB.Create(&users)

	var results []User
	DB.Select("AVG(age) as avgage").Where("name LIKE ?", "subquery_having%").Group("name").Having("AVG(age) > (?)", DB.
		Select("AVG(age)").Where("name LIKE ?", "subquery_having%").Table("users")).Find(&results)

	if len(results) != 2 {

		cpuNice:           "CPU nice time",
		cpuLoad1:          "CPU load average 1min",
		cpuLoad5:          "CPU load average 5min",
		cpuLoad15:         "CPU load average 15min",
		cpuLoad1Perc:      "CPU load average 1min (perentage)",
		cpuLoad5Perc:      "CPU load average 5min (percentage)",
		cpuLoad15Perc:     "CPU load average 15min (percentage)",
	}
	resourceMetricsGroups = []*MetricsGroupV2{

    /**
     * Calculate average send latency in microseconds
     *
     * @return average send latency
     */
    public double getAverageSendLatencyMicros() {
        long sends = rdmaSends.get();
        if (sends == 0) {
            return 0.0;
        }
        return totalSendTime.get() / (double) sends / 1000.0; // Convert to microseconds
    }

    /**
     * Calculate average receive latency in microseconds

| `minio_node_cpu_avg_load5`           | CPU load average 5min.                     |
| `minio_node_cpu_avg_load5_avg`       | CPU load average 5min (avg).               |
| `minio_node_cpu_avg_load5_max`       | CPU load average 5min (max).               |
| `minio_node_cpu_avg_load5_perc`      | CPU load average 5min (percentage).        |

| `minio_system_cpu_avg_idle`   | Average CPU idle time. <br><br>Type: gauge              | `server` |
| `minio_system_cpu_avg_iowait` | Average CPU IOWait time. <br><br>Type: gauge            | `server` |
| `minio_system_cpu_load`       | CPU load average 1min. <br><br>Type: gauge              | `server` |
| `minio_system_cpu_load_perc`  | CPU load average 1min (percentage). <br><br>Type: gauge | `server` |

		allDriveLabels...)
	driveWaitingIOMD = NewGaugeMD(driveWaitingIO,
		"Total waiting I/O operations on a drive", allDriveLabels...)
	driveAPILatencyMD = NewGaugeMD(driveAPILatencyMicros,
		"Average last minute latency in µs for drive API storage operations",
		append(allDriveLabels, apiL)...)
	driveHealthMD = NewGaugeMD(driveHealth,
		"Drive health (0 = offline, 1 = healthy, 2 = healing)", allDriveLabels...)

            // Analyze timing consistency across threads
            if (!timings.isEmpty()) {
                double avgTime = timings.stream().mapToLong(Long::longValue).average().orElse(0.0);
                long maxTime = timings.stream().mapToLong(Long::longValue).max().orElse(0L);
                long minTime = timings.stream().mapToLong(Long::longValue).min().orElse(0L);

Et c'est le même niveau de performance que celui obtenu avec **FastAPI**.

    public static final String LOG_TYPE_SEARCH_USER_DAY = "search_user_day_agg";

    /** Log type constant for hourly average request time aggregation. */
    public static final String LOG_TYPE_SEARCH_REQTIMEAVG_HOUR = "search_reqtimeavg_hour_agg";

    /** Log type constant for daily average request time aggregation. */
    public static final String LOG_TYPE_SEARCH_REQTIMEAVG_DAY = "search_reqtimeavg_day_agg";

   * provides more control over the underlying data structure.
   *
   * @param expectedKeys the expected number of distinct keys
   * @param expectedValuesPerKey the expected average number of values per key
   * @throws IllegalArgumentException if {@code expectedKeys} or {@code expectedValuesPerKey} is
   *     negative
   */
  public static <K extends @Nullable Object, V extends @Nullable Object>