Controlling where objects are populated into memory on RAC:

One of the things I looked at was the new FOR SERVICE sub-clause of the INMEMORY DISTRIBUTE clause.

ALTER TABLE customers INMEMORY PRIORITY HIGH DISTRIBUTE FOR SERVICE IM1;

This new sub-clause allows you to control exactly where an object (table or partition) is populated into memory in a RAC environment based on where a database service is allowed to run. If the service is stopped then the objects distributed for that service are automatically removed from the IM column store(s).

If you recall each RAC node has it’s own In-Memory column store (IM column store). By default, when a table is populated into memory in a RAC environment it will be automatically distributed across all of the IM column stores in the cluster. That is to say, a piece of the table will appear in each RAC node, effectively making the IM column store a shared-nothing architecture.

This is not always desirable, especially if you only run certain applications or workloads on a subset of RAC node.

-- Step1 configure the database services

srvctl ADD service -db orcl -service IM1 –preferred “rac1”

 srvctl ADD service -db orcl -service IM2 -r preferred “rac2”

 srvctl ADD service -db orcl -service IM_ALL -r preferred “rac1,rac2”

 srvctl START service -db orcl -service "IM1,IM2,IM_ALL" 

srvctl STATUS service -db orcl

 -- Step 2 Add INMEMORY attributes to the SALES, CUSTOMERS and PRODUCTS tables

 ALTER TABLE customers INMEMORY PRIORITY HIGH DISTRIBUTE FOR SERVICE IM1;

 ALTER TABLE products INMEMORY PRIORITY MEDIMUM DISTRIBUTE FOR SERVICE IM2;

 ALTER TABLE sales INMEMORY PRIOIRTY HIGH;

 -- Step 3 Check the content of the In-Memory column store

 SELECT * FROM   gv$inmemory_area;

 SELECT v.inst_id,  v.owner, v.segment_name name, v.populate_status, SUM(v.bytes_not_populated)/1024/1024 MB_not_populated FROM   gv$im_segments v GROUP  BY v.inst_id,  v.owner, v.segment_name, v.populate_status; 

 SELECT m.inst_id, m.blocksinmem,  m.datablocks FROM   gv$im_segments_detail m,  user_objects o WHERE  m.dataobj = o.object_id AND    o.object_name = 'SALES'; 

 -- Step 4 create a new session

-- Run query and capture the session level statistics before and after the query

 SELECT t1.name, t2.value FROM v$sysstat t1, v$mystat t2 WHERE t1.name IN ('table scans (long tables)','table scans (IM)', 'session logical reads', 'session logical reads - IM', 'IM scan rows',  'IM scan segments disk') AND t1.statistic# = t2.statistic# ORDER BY t1.name;

 SELECT c.cust_city,  p.prod_name, SUM(s.amount_sold) FROM   sales s, customers c, products p WHERE  s.cust_id = c.cust_id AND    s.prod_id = p.prod_id GROUP  BY c.cust_city, p.prod_name; 

 SELECT t1.name, t2.value FROM v$sysstat t1, v$mystat t2 WHERE t1.name IN ('table scans (long tables)','table scans (IM)', 'session logical reads', 'session logical reads - IM', 'IM scan rows',  'IM scan segments disk')AND t1.statistic# = t2.statistic# ORDER BY t1.name;

 -- Step 5 Repopulate the tables so they are all on the same node

 ALTER TABLE products INMEMORY PRIORITY MEDIMUM DISTRIBUTE FOR SERVICE IM1;

 ALTER TABLE sales INMEMORY PRIOIRTY HIGH DISTRIBUTE FOR SERVICE IM1;

 -- Step 6 Check the content of the In-Memory column store

 SELECT v.inst_id,  v.owner, v.segment_name name, v.populate_status,  SUM(v.bytes_not_populated)/1024/1024 MB_not_populated FROM   gv$im_segments v GROUP  BY v.inst_id,  v.owner, v.segment_name, v.populate_status; 

 -- Step 7 Create a new session and rerun the query and check the session statistics before and after 

 SELECT t1.name, t2.value FROM v$sysstat t1, v$mystat t2 WHERE t1.name IN ('table scans (long tables)','table scans (IM)', 'session logical reads', 'session logical reads - IM', 'IM scan rows', 'IM scan segments disk')AND t1.statistic# = t2.statistic# ORDER BY t1.name;

 SELECT c.cust_city, p.prod_name,  SUM(s.amount_sold) FROM   sales s,  customers c,  products p WHERE  s.cust_id = c.cust_id AND    s.prod_id = p.prod_id GROUP  BY c.cust_city, p.prod_name; 

 SELECT t1.name, t2.value FROM v$sysstat t1, v$mystat t2 WHERE t1.name IN ('table scans (long tables)', 'table scans (IM)', 'session logical reads', 'session logical reads - IM', 'IM scan rows',  'IM scan segments disk')AND t1.statistic# = t2.statistic# ORDER BY t1.name;

 -- Step 8 stop the service and check that all of the objects were removed from the IM column store

 srvctl stop service -db main -service "IM1"

 SELECTv.inst_id, v.owner,  v.segment_namename, v.populate_status,  SUM(v.bytes_not_populated)/1024/1024 MB_not_populated FROM   gv$im_segments v GROUP  BY v.inst_id,  v.owner, v.segment_name, v.populate_status;