collect schema level metrics (#139)

Author: maskarb

kokudaily/sql/key_metrics/cust_cloud_costs_data.sql

@@ -1,4 +1,5 @@
 select
+    row_number() OVER () AS id, -- exclude schema to anonymize data
     to_char(DATE_TRUNC('month', date), 'YYYY-MM') AS month,
     SUM(aws_unblended_cost) AS "aws_unblended_cost",
     SUM(aws_calculated_amortized_cost) AS "aws_calculated_amortized_cost",
@@ -7,6 +8,6 @@ select
     SUM(gcp_total) AS "gcp_total",
     SUM(oci_cost) AS "oci_cost"
 from __cust_cloud_cost_report
-GROUP BY DATE_TRUNC('month', date)
-ORDER BY DATE_TRUNC('month', date)
+GROUP BY schema, DATE_TRUNC('month', date)
+ORDER BY schema, DATE_TRUNC('month', date)
 ;

kokudaily/sql/key_metrics/cust_cloud_costs_data_setup.sql

@@ -2,7 +2,7 @@ DROP TABLE IF EXISTS __cust_cloud_cost_report;
 
 -- create temp table for results
 CREATE TEMPORARY TABLE IF NOT EXISTS __cust_cloud_cost_report (
-    id serial,
+    schema text,
     date date,
     aws_unblended_cost numeric(33, 2),
     aws_calculated_amortized_cost numeric(33, 2),
@@ -18,6 +18,7 @@ DECLARE
     schema_rec record;
     stmt_tmpl text = '
 INSERT INTO __cust_cloud_cost_report (
+    schema,
     date,
     aws_unblended_cost,
     aws_calculated_amortized_cost,
@@ -119,7 +120,7 @@ oci_costs AS (
     GROUP BY date
 )
 SELECT
-    -- awc.customer AS "customer", -- customer is used for grouping, but left off report for anonymity
+    awc.customer AS "schema",
     date AS "date",
     awc.aws_unblended_cost AS "aws_unblended_cost",
     awc.aws_calculated_amortized_cost AS "aws_calculated_amortized_cost",

kokudaily/sql/key_metrics/cust_openshift_costs_data.sql

@@ -1,4 +1,5 @@
 select
+    row_number() OVER () AS id, -- exclude schema to anonymize data
     to_char(DATE_TRUNC('month', date), 'YYYY-MM') AS month,
     SUM(total_infrastructure_raw_cost) AS "total_infrastructure_raw_cost",
     SUM(total_cost_model_costs) AS "total_cost_model_costs",
@@ -11,6 +12,6 @@ select
     SUM(sup_cost_model_memory_cost) AS "sup_cost_model_memory_cost",
     SUM(sup_cost_model_volume_cost) AS "sup_cost_model_volume_cost"
 from __cust_openshift_cost_report
-GROUP BY DATE_TRUNC('month', date)
-ORDER BY DATE_TRUNC('month', date)
+GROUP BY schema, DATE_TRUNC('month', date)
+ORDER BY schema, DATE_TRUNC('month', date)
 ;

kokudaily/sql/key_metrics/cust_openshift_costs_data_setup.sql

@@ -2,7 +2,7 @@ DROP TABLE IF EXISTS __cust_openshift_cost_report;
 
 -- create temp table for results
 CREATE TEMPORARY TABLE IF NOT EXISTS __cust_openshift_cost_report (
-    id serial,
+    schema text,
     date date,
     total_infrastructure_raw_cost numeric(33, 2),
     total_cost_model_costs numeric(33, 2),
@@ -22,6 +22,7 @@ DECLARE
     schema_rec record;
     stmt_tmpl text = '
 INSERT INTO __cust_openshift_cost_report (
+    schema,
     date,
     total_infrastructure_raw_cost,
     total_cost_model_costs,
@@ -121,7 +122,7 @@ sup_costs AS (
     GROUP BY date
 )
 SELECT
-    -- ir.customer AS "customer", -- customer is used for grouping, but left off report for anonymity
+    ir.customer AS "schema",
     date AS "date",
     COALESCE(ir.infrastructure_raw_cost, 0) AS "total_infrastructure_raw_cost",
     ic.infra_total_cost_model+sc.sup_total_cost_model AS "total_cost_model_costs",

kokudaily/sql/key_metrics/cust_openshift_infra_data.sql

@@ -1,26 +1,27 @@
 SELECT
+    row_number() OVER () AS id, -- exclude schema to anonymize data
     to_char(DATE_TRUNC('month', date), 'YYYY-MM') AS month,
-    SUM(cluster_count) AS total_cluster_count,
-    SUM(node_count) AS total_node_count,
-    SUM(infra_node_count) AS total_infra_node_count,
-    SUM(control_plane_node_count) AS total_control_plane_node_count,
-    SUM(worker_node_count) AS total_worker_node_count,
-    SUM(infra_node_cpu_cores) AS total_infra_node_cpu_cores,
-    SUM(control_plane_node_cpu_cores) AS total_control_plane_node_cpu_cores,
-    SUM(worker_node_cpu_cores) AS total_worker_node_cpu_cores,
-    SUM(infra_node_mem_gb) AS total_infra_node_mem_gb,
-    SUM(control_plane_node_mem_gb) AS total_control_plane_node_mem_gb,
-    SUM(worker_node_mem_gb) AS total_worker_node_mem_gb,
-    SUM(pvc_count) AS total_pvc_count,
-    SUM(cluster_capacity_cores) AS total_cluster_capacity_cores,
+    MAX(cluster_count) AS total_cluster_count,
+    MAX(node_count) AS total_node_count,
+    MAX(infra_node_count) AS total_infra_node_count,
+    MAX(control_plane_node_count) AS total_control_plane_node_count,
+    MAX(worker_node_count) AS total_worker_node_count,
+    MAX(infra_node_cpu_cores) AS total_infra_node_cpu_cores,
+    MAX(control_plane_node_cpu_cores) AS total_control_plane_node_cpu_cores,
+    MAX(worker_node_cpu_cores) AS total_worker_node_cpu_cores,
+    MAX(infra_node_mem_gb) AS total_infra_node_mem_gb,
+    MAX(control_plane_node_mem_gb) AS total_control_plane_node_mem_gb,
+    MAX(worker_node_mem_gb) AS total_worker_node_mem_gb,
+    MAX(pvc_count) AS total_pvc_count,
+    MAX(cluster_capacity_cores) AS total_cluster_capacity_cores,
     SUM(cluster_capacity_core_hours) AS total_cluster_capacity_core_hours,
-    SUM(cluster_capacity_memory_gb) AS total_cluster_capacity_memory_gb,
+    MAX(cluster_capacity_memory_gb) AS total_cluster_capacity_memory_gb,
     SUM(cluster_capacity_memory_gb_hours) AS total_cluster_capacity_memory_gb_hours,
-    SUM(volume_request_gb) AS total_volume_request_gb,
+    MAX(volume_request_gb) AS total_volume_request_gb,
     SUM(volume_request_gb_mo) AS total_volume_request_gb_mo,
-    SUM(pvc_capacity_gb) AS total_pvc_capacity_gb,
+    MAX(pvc_capacity_gb) AS total_pvc_capacity_gb,
     SUM(pvc_capacity_gb_mo) AS total_pvc_capacity_gb_mo
 FROM __cust_openshift_infra_report
-GROUP BY DATE_TRUNC('month', date)
-ORDER BY DATE_TRUNC('month', date)
+GROUP BY schema, DATE_TRUNC('month', date)
+ORDER BY schema, DATE_TRUNC('month', date)
 ;

kokudaily/sql/key_metrics/cust_openshift_infra_data_setup.sql

@@ -2,7 +2,7 @@ DROP TABLE IF EXISTS __cust_openshift_infra_report;
 
 -- create temp table for results
 CREATE TEMPORARY TABLE IF NOT EXISTS __cust_openshift_infra_report (
-    id serial,
+    schema text,
     date date,
     cluster_count integer,
     node_count integer,
@@ -32,6 +32,7 @@ DECLARE
     schema_rec record;
     stmt_tmpl text = '
 INSERT INTO __cust_openshift_infra_report (
+    schema,
     date,
     cluster_count,
     node_count,
@@ -161,7 +162,7 @@ storage_agg AS (
     GROUP BY date
 )
 SELECT
-    -- customer is used for grouping, but left off report for anonymity
+    customer AS "schema",
     date,
     cluster_count,
     node_count,