Visualizing API Response Times with Python Plotly Violin Charts

During routine API testing the author stored each request's latency in a MySQL database, accumulating a sizable dataset. To make the data easier to interpret, they learned Python with Plotly and created a violin plot that visualizes the response‑time distribution for several API endpoints.

The main script performs the following steps:

#!/usr/bin/python
# coding=utf-8
import plotly.plotly
import pandas as pd
import plotly.figure_factory as ff
import second.mysql

if __name__ == "__main__":
    a = second.mysql.Mysql()
    b = a.getApiTimes('/article/list/userfeed', '/article/detail', '/article/info', '/advertise/api/list', '/common/menu')
    df = pd.DataFrame(dict(Score=b[0], Group=b[1]))  # combine data
    fig = ff.create_violin(df, data_header='Score', group_header='Group',
                           height=700, width=1200, title='接口请求时间')
    plotly.offline.plot(fig)

The getApiTimes method defined in the MySQL helper class queries the api_result table for each supplied endpoint, filters out abnormal records (response time < 1 second), and returns two parallel lists: the recorded times and the corresponding API names.

def getApiTimes(self, *params):
    num = str(params.__len__())
    print "接口数：" + num
    conn = self.conn
    cur = conn.cursor()
    data = []
    size = []
    for api in params:
        cur.execute("SELECT * FROM api_result WHERE api_name = \"" + api + "\"")
        dfs = cur.fetchall()
        for row in dfs:
            if row[7] < 1:
                data.append(row[7])
                size.append(api)
    cur.close()
    conn.commit()
    conn.close()
    return [data, size]

The resulting violin chart (shown below) makes the distribution of API latencies immediately apparent, providing a clear visual aid for presentations or performance reports.

This approach demonstrates a practical workflow for turning raw performance metrics into insightful visualizations without requiring heavyweight BI tools.