Upload
trinhanh
View
213
Download
0
Embed Size (px)
Citation preview
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
In-Memory is Your Data
Warehouse’s New BFF
Michelle Kolbe
@MeKolbe linkedin.com/in/michelle.kolbemedium.com/@datacheesehead
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
Red Pill Analytics: Leadership
3
Stewart Bryson
Oracle ACE Director
Previously From:
Rittman Mead
Informix Software
Data Warehousing &
Business Intelligence
since 1996
Kevin McGinley
Oracle ACE Director
Previously From:
Accenture
BI Consulting Group
Data Warehousing &
Business Intelligence
since 1997
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
About Me
•
••
•
•
•
•
•
•
A unique approach to BI
and analytics delivery.
www.redpillanalytics.com
@RedPillA
•Development-as-a-Service
•Continuous Integration Products
•Agile Delivery
•Subscription Billing
•Traditional Consulting and Training
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
Agenda
•
•
•
•
•12.2 New Features for In-Memory
7
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
Breakthrough: Dual-Format Database
•
•
•
•
9
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
Configuring the In-Memory Column Store
•
•
•
•
•
10
© 2014 RED PILL Analytics
IMCU pool: In Memory Compression Units
Actual Columnar Data
SMU pool: Snapshot Metadata Units(SMUs)
Metadata about Columnar Data
Columnar representation of many rows
Rows from one or more table extents
Dependent on row, compression factor, etc.
© 2017 RED PILL Analytics
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
In-Memory Compression Unit
•
•
•
•
•
•
17
Not Persisted on Storage
2x to 20x Compression
Enabled at the Segment Level
© 2014 RED PILL Analytics
ALTER TABLE sales INMEMORY;
ALTER TABLE sales NO INMEMORY;
ALTER TABLE sales INMEMORY
NO INMEMORY (PROD_ID);
CREATE TABLE customers ……
PARTITION BY LIST
(PARTITION p1 …… INMEMORY,
(PARTITION p2 …… NO INMEMORY);
Recap
Size of the column store is controlled by
inmemory_size parameter
alter table lineorder inmemory;
Monitor the column store using v$im_segments
v$im_segments also shows our compress ratios
Why is the In-Memory Column Store faster than the
Buffer Cache?
It’s all memory after all.
© 2017 RED PILL Analytics
Recap
Alter session set inmemory_query=disable;
inmemory & no_inmemory hints
Column store 40X faster than buffer cache on
single row access
Storage index MIN/MAX pruning reduces data
scans significantly
www.RedPillAnalytics.com [email protected] @RedPillA © 2017 RED PILL Analytics
In-Memory is Your Data
Warehouse’s New BFF
Michelle Kolbe
@MeKolbe linkedin.com/in/michelle.kolbemedium.com/@datacheesehead