group N by column1,column2,column3

General Discussions - English
1

hi！I meet a question. When I use the (group N by column1, column2 … ) statement, manticore search speed is slow. as follow：

select id, content_type, protocol, request_type from testplain group 2 by content_type, protocol,request_type order by protocol ASC limit 1000

OK
时间: 54.638s

I tried running (group by cloumn1,column2…)

select id, content_type, protocol, request_type from testplain group by content_type, protocol,request_type order by protocol ASC limit 1000

OK
时间: 2.223s

How can I improve the search speed of (group by N) statement

Briefly introduction of my index:
10G,7500000 documents, 48 columns
content_type,protocol columns ： random three-digit number
request_type column: random 1-digit number

2

Hi @ordinary

The difference seems to be too big and it may be a bug. Could you file an issue on github Issues · manticoresoftware/manticoresearch · GitHub and upload your index to our FTP - Manticore Search Manual ?

3

Thanks for your replay！
I have submitted my question on github and the indexes file has been uploaded via ftp.