Authors:
Niloy Mukherjee
;
Kartik Kulkarni
;
Hui Jin
;
Jesse Kamp
and
Tirthankar Lahiri
Affiliation:
Oracle Corporation, United States
Keyword(s):
Oracle RDBMS in-Memory Option, Dual-format Distributed in-Memory Database, Scale-out, in-Memory
Compression Units (IMCUs), Automated Distribution, Distributed SQL Execution.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Business Analytics
;
Cardiovascular Technologies
;
Computing and Telecommunications in Cardiology
;
Data Engineering
;
Decision Support Systems
;
Decision Support Systems, Remote Data Analysis
;
Distributed and Mobile Software Systems
;
Distributed Architectures
;
Health Engineering and Technology Applications
;
Knowledge-Based Systems
;
Parallel and High Performance Computing
;
Software Engineering
;
Symbolic Systems
Abstract:
The Oracle RDBMS In-memory Option (DBIM), introduced in 2014, is an industry-first distributed dual format in-memory RDBMS that allows a database object to be stored in columnar format purely in-memory, simultaneously maintaining transactional consistency with the corresponding row-major format persisted in storage and accessed through in-memory database buffer cache. The in-memory columnar format is highly optimized to break performance barriers in analytic query workloads while the row format is most suitable for OLTP workloads. In this paper, we present the distributed architecture of the Oracle Database In- memory Option that enables the in-memory RDBMS to transparently scale out across a set of Oracle database server instances in an Oracle RAC cluster, both in terms of memory capacity and query processing throughput. The architecture allows complete application-transparent, extremely scalable and automated in- memory distribution of Oracle RDBMS objects across multiple instances
in a cluster. It seamlessly provides distribution awareness to the Oracle SQL execution framework, ensuring completely local memory scans through affinitized fault-tolerant parallel execution within and across servers without explicit optimizer plan changes or query rewrites.
(More)