Create group based on the data available in SQL server table having specific condition? - sql

Question

There is a table which contains SQL server blocking chain data, like below.
I am trying to pull only those blocking chain groups whose average wait time is greater than 20 seconds.Group can be identified like - It starts from where it founds blocked value as 0 and ends on where it found again blocked value as 0. And last found with 0 value should not be consider in group
Blocking_time SPID blocked WAIT_MS Blocking_Chain_tree_details_by_Session_id_and_header Wait_type
7/28/19 5:14 AM 130 0   HEAD -  SPID (130) - EL.dbo.test;1
7/28/19 5:14 AM 292 130 1   |      |-----  SPID (292) - EL.dbo.test123;1 PAGELATCH_EX
7/28/19 5:14 AM 949 130 1   |      |-----  SPID (949) - EL.dbo.sstest123;1 PAGELATCH_EX
7/28/19 5:32 AM 106 130 1   |      |-----  SPID (106) - EL.dbo.checjmark;1 PAGELATCH_EX
7/28/19 5:32 AM 130 0   HEAD -  SPID (130) - Eli.dbo.sss;1
7/28/19 5:32 AM 292 130 1   |      |-----  SPID (292) - EL.dbo.variable;1 PAGELATCH_EX
7/28/19 5:32 AM 949 130 1   |      |-----  SPID (949) - Eldbo.anything;1 PAGELATCH_EX
7/28/19 5:32 AM 1578 130 12000   |      |-----  SPID (1578) - EL.dbo.something;1 PAGELATCH_EX
7/28/19 9:20 AM 196 513 21700   |      |-----  SPID (196) - (#P1 uniqueidentifier,#P2 int,#P3 int,#P4 int,#P5 int,#P6 int,#P7 int,#P8 int,#P ... LCK_M_IX
NA
Actual result should be like as below-
Blocking_time SPID blocked WAIT_MS Blocking_Chain_tree_details_by_Session_id_and_header Wait_type
7/28/19 5:32 AM 130 0   HEAD -  SPID (130) - Eli.dbo.sss;1
7/28/19 5:32 AM 292 130 1   |      |-----  SPID (292) - EL.dbo.variable;1 PAGELATCH_EX
7/28/19 5:32 AM 949 130 1   |      |-----  SPID (949) - Eldbo.anything;1 PAGELATCH_EX
7/28/19 5:32 AM 1578 130 12000   |      |-----  SPID (1578) - EL.dbo.something;1 PAGELATCH_EX
7/28/19 9:20 AM 196 513 21700   |      |-----  SPID (196) - (#P1 uniqueidentifier,#P2 int,#P3 int,#P4 int,#P5 int,#P6 int,#P7 int,#P8 int,#P ... LCK_M_IX

Answer 1

You can use a window function for this. So long as you put your grouping columns in the PARTITION BY you'll be able to get the MAX value for the group. Then you can filter to just the groups where the max time is over 20 seconds.
SELECT *
FROM
(
SELECT Blocking_time,
SPID,
blocked,
WAIT_MS,
Blocking_Chain_tree_details_by_Session_id_and_header,
Wait_type,
MAX(WAIT_MS) OVER (PARTITION BY Blocking_time ORDER BY Blocking_time ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) [Max_WAIT_MS]
FROM <YourTable>
) rawData
WHERE [Max_WAIT_MS] > 20000