I'm using AWS Athena to parse my Application Load Balancer logs.
I'm trying to get a list of browsers and for each browsers, the number of unique user.
I've managed to get this list but the user count is not correct. I don't know how to group users by their IP.
1 Google Chrome 9000000
2 Apple Safari 8000000
3 Unknown 5000000
4 Mozilla Firefox 2000000
5 Internet Explorer 10000
6 Outlook 10000
7 Opera 88
8 Edge 7
Here is the query
SELECT DISTINCT
CASE
WHEN user_agent LIKE '%edge%'THEN 'Edge'
WHEN user_agent LIKE '%MSIE%' THEN
'Internet Explorer'
WHEN user_agent LIKE '%Firefox%' THEN
'Mozilla Firefox'
WHEN user_agent LIKE '%Chrome%' THEN
'Google Chrome'
WHEN user_agent LIKE '%Safari%' THEN
'Apple Safari'
WHEN user_agent LIKE '%Opera%' THEN
'Opera'
WHEN user_agent LIKE '%Outlook%' THEN
'Outlook'
ELSE 'Unknown'
END AS browser , COUNT(client_ip) AS Number
FROM alb_logs
WHERE parse_datetime(time,'yyyy-MM-DD''T''HH:mm:ss.SSSSSS''Z')
BETWEEN parse_datetime('2018-01-01-00:00:00','yyyy-MM-DD-HH:mm:ss')
AND parse_datetime('2018-07-18-00:00:00','yyyy-MM-DD-HH:mm:ss')
GROUP BY CASE
WHEN user_agent LIKE '%edge%'THEN 'Edge'
WHEN user_agent LIKE '%MSIE%' THEN
'Internet Explorer'
WHEN user_agent LIKE '%Firefox%' THEN
'Mozilla Firefox'
WHEN user_agent LIKE '%Chrome%' THEN
'Google Chrome'
WHEN user_agent LIKE '%Safari%' THEN
'Apple Safari'
WHEN user_agent LIKE '%Opera%' THEN
'Opera'
WHEN user_agent LIKE '%Outlook%' THEN
'Outlook'
ELSE 'Unknown'
END
ORDER BY Number DESC
I'm missing some kind of group by client_ip
, but the result would be wrong...
You need COUNT(DISTINCT client_ip)
aggregation and also you don't need SELECT DISTINCT
, like this
SELECT CASE WHEN user_agent ... END AS browser, COUNT(DISTINCT client_ip) AS Number
FROM alb_logs
WHERE ...
GROUP BY 1
ORDER BY 2 DESC