Oracle TimesTen: An In-Memory

Database for Enterprise Applications

A Paper by Tirthankar Lahiri, Marie-Anne Neimat & Steve Folkman

Presented by

Anchalin Puangmaha Rungnapa Ruangthananrak

MA561: Seminar in Accounting Information Systems

Conclusion Experimental

Results IMDB Cache

Grid High

Availability Basic

Architecture Introduction

TimesTen is an Oracle in-Memory database product

Experimental Result

IMDB Cache Grid

High Availability

Basic Architecture

Conclusions and

FutureWork

Experimental Results

IMDB Cache Grid

High Availability Introduction

Basic Architecture

Basic Architecture : 1. APIs and connectivity

2. Basic storage manager

design

3. Transactional durability

Conclusions and

FutureWork

Experimental Results

IMDB Cache Grid Introduction

Basic Architecture

High Availability

If a failure of the Active occurs, the Standby can be promoted to become an Active Database. If a failure of the Standby occurs, the Active will then replicate directly to the Read Only Subscribers.

Conclusions and

FutureWork

Experimental Results Introduction

Basic Architecture

High Availability

IMDB Cache Grid

Cache groups is a syntactic declaration and comprises one Root table, and optional Child tables related by foreign key constraints.

Root Table

Child tables

In-memory cache grid, which consists of a collection of IMDB Caches for an application’s cached data and is deployed in the application tier. Cached data is distributed among the grid members.

Conclusions and

FutureWork Introduction

Basic Architecture

High Availability

IMDB Cache Grid

Experimental Results

Response time and throughput : The combination of in-memory execution and direct-linked access allow TimesTen to provide extremely low response times

The experiment of response times was run on a standard commodity Intel Xeon 5670- system with 2 processors and 12 cores, running Oracle Enterprise Linux 5.6. The measurement of throughput were taken on a SPARC T5-8 system with 128 cores for TPTBM* and on a SPARC T5-2 system with 32 cores for a mobile call processing benchmark.

* TPTBM : The standard TimesTen throughput performance benchmark

Conclusions and

FutureWork Introduction

Basic Architecture

High Availability

IMDB Cache Grid

Experimental Results

(a) Lookup on primary key (b) Table scan

Scaling across a grid : how TimesTen scales out when extended to the IMDB cache configuration [Left Graph] the grid scales very effectively as we increase the number of grid members [Right Graph] the performance of global scans scales fairly linearly with the number of participating grid members

Introduction Basic

Architecture High

Availability IMDB Cache

Grid Experimental

Results Conclusion

Conclusions

ACID properties and transactional semantics

Storage manager optimized for IMDB and for high concurrency

SQL support with extensions for analytics, stored procedures and access

Very low response times as well as very good scaling characteristics

Full database as well as table-level multi-master replication for high availability

List Of Capabilities

Thank You