indexing using Utilization dataset

Author: ptyadana

Advanced SQL for Data Science - Time Series/02.03.Modeling Time Series Data - Indexing - Utilization dataset.sql

@@ -0,0 +1,51 @@
+/*********** Partitioning Data (using Utilization data set) ****************/
+
+-- Load data from text file
+COPY time_series.utilization(
+	event_time,server_id, cpu_utilization, free_memory, session_cnt
+) FROM 'C:\temp-sql\utilization.txt' DELIMITER ',';
+
+
+-- check data
+SELECT * FROM time_series.utilization
+LIMIT 5;
+
+
+
+/* average CPU utilization by server id : cost - 3687.71*/
+EXPLAIN SELECT server_id, AVG(cpu_utilization)
+FROM time_series.utilization
+WHERE event_time BETWEEN '2019-03-05' AND '2019-03-06'
+GROUP BY server_id;
+
+
+------------ we will create index to speed things up ---------------
+
+-- 1) index of event time, server id
+CREATE INDEX idx_event_time_utilization
+ON time_series.utilization(event_time, server_id);
+
+/* after indexing cost - 3651.71*/
+EXPLAIN SELECT server_id, AVG(cpu_utilization)
+FROM time_series.utilization
+WHERE event_time BETWEEN '2019-03-05' AND '2019-03-06'
+GROUP BY server_id;
+
+DROP INDEX time_series.idx_event_time_utilization;
+
+
+-- 2) index of server id, event time (switch order)
+CREATE INDEX idx_server_event_utilization
+ON time_series.utilization(server_id, event_time);
+
+
+/* 
+now the index is not even used - 3687.71
+
+Notice that when event time is in second column to be indexed. 
+So with only index being server_id as first then by event_time, we wouldn't able to use index at all.
+*/
+EXPLAIN SELECT server_id, AVG(cpu_utilization)
+FROM time_series.utilization
+WHERE event_time BETWEEN '2019-03-05' AND '2019-03-06'
+GROUP BY server_id;