Skip Headers
Oracle® In-Memory Database Cache Introduction
Release 11.2.1

Part Number E14261-06
Go to Documentation Home
Home
Go to Book List
Book List
Go to Table of Contents
Contents
Go to Index
Index
Go to Master Index
Master Index
Go to Feedback page
Contact Us

Go to previous page
Previous
Go to next page
Next
View PDF

1 Oracle TimesTen In-Memory Database and Oracle In-Memory Database Cache

Oracle TimesTen In-Memory Database (TimesTen) is a memory-optimized relational database that empowers applications with the responsiveness and high throughput required by today's real-time enterprises and industries such as telecom, capital markets and defense. Oracle In-Memory Database Cache (IMDB Cache) uses the Oracle TimesTen In-Memory Database as its RDBMS engine. Deployed in the application tier as an embedded database, Oracle TimesTen In-Memory Database operates on databases that fit entirely in physical memory using standard SQL interfaces. High availability for the in-memory database is provided through real-time transactional replication.

Oracle In-Memory Database Cache (IMDB Cache) is an Oracle Database product option ideal for caching performance-critical subsets of an Oracle database for improved response time in the application tier. Cache tables can be read-only or updatable. Applications read and update the cache tables using standard SQL, and data synchronization between the cache and the Oracle Database is performed automatically. Oracle In-Memory Database Cache offers applications the full generality and functionality of a relational database, the transparent maintenance of cache consistency with the Oracle Database, and the real-time performance of an in-memory database.

Oracle TimesTen In-Memory Database and IMDB Cache deliver real-time performance by changing the assumptions about where data resides at runtime. By managing data in memory and optimizing data structures and access algorithms accordingly, database operations execute with maximum efficiency, achieving dramatic gains in responsiveness and throughput, even compared with a fully cached, disk-based relational database management system (RDBMS). TimesTen and IMDB Cache libraries are also embedded within applications, eliminating context switching and unnecessary network operations, further improving performance.

Following the standard relational data model, you can use SQL, JDBC, ODBC, PL/SQL and Oracle Call Interface (OCI) to access TimesTen and IMDB Cache databases. Using SQL to shield applications from system internals allows databases to be altered or extended without impacting existing applications. New services can be quickly added into a production environment simply by adding application modules, tables and columns. As with any mainstream RDBMS, a cost-based optimizer automatically determines the fastest way to process queries and transactions. Any developer familiar with the Oracle Database or SQL interfaces can be immediately productive developing real-time applications with TimesTen and IMDB Cache.

Why is Oracle TimesTen In-Memory Database fast?

Much of the work that is done by a conventional, disk-optimized RDBMS is done under the assumption that data primarily resides on disk. Optimization algorithms, buffer pool management, and indexed retrieval techniques are designed based on this fundamental assumption.

Even when a disk-based RDBMS has been configured to hold all of its data in main memory, its performance is hobbled by assumptions of disk-based data residency. These assumptions cannot be easily reversed because they are hard-coded in processing logic, indexing schemes, and data access mechanisms.

TimesTen is designed with the knowledge that data resides in main memory and can take more direct routes to data, reducing the length of the code path and simplifying algorithms and structure.

When the assumption of disk-residency is removed, complexity is dramatically reduced. The number of machine instructions drops, buffer pool management disappears, extra data copies are not needed, index pages shrink, and their structure is simplified. The design becomes simple and more compact, and requests are executed faster. Figure 1-1 shows the simplicity of the TimesTen design.

Figure 1-1 Comparing a disk-based RDBMS to TimesTen

Description of Figure 1-1 follows
Description of "Figure 1-1 Comparing a disk-based RDBMS to TimesTen"

In a conventional disk-based RDBMS, client applications communicate with a database server process over some type of IPC connection, which adds performance overhead to all SQL operations. An application can link TimesTen directly into its address space to eliminate the IPC overhead and streamline query processing. This is accomplished through a direct connection to TimesTen. Traditional client/server access is also supported for functions such as reporting, or when a large number of application-tier platforms must share access to a common in-memory database. From an application's perspective, the TimesTen API is identical whether it is a direct connection or a client/server connection.

TimesTen and IMDB Cache feature overview

TimesTen and IMDB Cache have many familiar database features as well as some unique features. This section includes the following topics:

ODBC and JDBC interfaces

TimesTen and IMDB Cache support ODBC and JDBC. Unlike many other database systems, where ODBC or JDBC API support may be much slower than the proprietary interface, ODBC and JDBC are native TimesTen interfaces that operate directly with the database engine. TimesTen supports versions of these APIs that are both fully compliant with the standards and tuned for maximum performance in the TimesTen environment.

SQL and PL/SQL functionality

TimesTen and IMDB Cache support extensive SQL functionality as well as SQL extensions to simplify the configuration and management of special features such as replication and IMDB Cache.

TimesTen and IMDB Cache support PL/SQL (Procedural Language Extension to SQL), a programming language that enables you to integrate procedural constructs with SQL for a TimesTen or IMDB Cache database.

OCI and Pro*C/C++ Precompiler support

TimesTen and IMDB Cache support the Oracle Call Interface (OCI) and the Pro*C/C++ Precompiler for TimesTen functionality.

TimesTen OCI depends on the Oracle client library and the TimesTen ODBC libraries. TimesTen OCI support enables you to run many existing OCI applications with TimesTen in direct mode or client/server mode. It also enables you to use other Oracle products, such as Pro*C/C++, that use OCI as a database interface. You can also call PL/SQL from OCI applications.

You can use the Pro*C/C++ Precompiler with embedded SQL and PL/SQL applications that access a TimesTen database.

For more information, see Oracle TimesTen In-Memory Database C Developer's Guide.

Access Control

TimesTen and IMDB Cache are installed with access control to allow only users with specific privileges to access particular TimesTen features. TimesTen Access Control uses standard SQL operations to establish user accounts with specific privileges. TimesTen supports TimesTen offers object-level access control as well as database-level access control.

For more information, see Oracle TimesTen In-Memory Database Operations Guide.

Distributed transactions

TimesTen and IMDB Cache support distributed transactions through the XA and JTA interfaces. These standard interfaces allow TimesTen to interoperate with transaction managers in distributed transaction processing (DTP) environments.

Database connectivity

TimesTen and IMDB Cache support direct driver connections for higher performance, as well as connections through a driver manager. TimesTen also supports client/server connections.

These connection options allow users to choose the best tradeoff between performance and functionality for their applications. Direct driver connections are fastest. Client/server connections may provide more flexibility. Driver manager connections can provide support for ODBC applications written for a different ODBC version or for multiple RDBMS products with ODBC interfaces.

See "TimesTen connection options".

Durability

TimesTen and IMDB Cache achieve durability through a combination of transaction logging and periodic refreshes of a disk-resident version of the database. Log records are written to disk asynchronously or synchronously to the completion of the transaction and controlled by the application at the transaction level. For systems where maximum throughput is paramount, such as non-monetary transactions within network systems, asynchronous logging allows extremely high throughput with minimal exposure. In cases where data integrity must be preserved, such as securities trading, TimesTen and IMDB Cache ensure complete durability, with no loss of data.

TimesTen uses the transaction log in the following situations:

  • Recover transactions if the application or database fails

  • Undo transactions that are rolled back

  • Replicate changes to other TimesTen databases

  • Replicate TimesTen changes to Oracle tables

  • Enable applications to detect changes to tables (using the XLA API)

TimesTen and IMDB Cache maintain the disk-resident version of the database with a checkpoint operation that takes place in the background and has very little impact on database applications. This operation is called a "fuzzy" checkpoint and is performed automatically. TimesTen and IMDB Cache also have a blocking checkpoint that does not require transaction log files for recovery. Blocking checkpoints must be initiated by the application. TimesTen and IMDB Cache maintain two checkpoint files in case a failure occurs mid-checkpoint. Checkpoint files should reside on disks separate from the transaction logs to minimize the impact of checkpointing on application activity.

See the following sections for more information:

Query optimization

TimesTen and IMDB Cache have a cost-based query optimizer that chooses the best query plan based on factors such as the presence of indexes, the cardinality of tables and the presence of ORDER BY clauses in the query.

Optimizer cost sensitivity is somewhat higher in TimesTen and IMDB Cache than in disk-based systems because the cost structure of a main-memory system differs from that of disk-based systems in which disk access is a dominant cost factor. Because disk access is not a factor in TimesTen and IMDB Cache, the optimization cost model includes factors not usually considered by optimizers for disk-based systems, such as the cost of evaluating predicates.

TimesTen and IMDB Cache provide range, hash and bitmap indexes and support two types of join methods (nested-loop and merge-join). The optimizer can create temporary indexes as needed.

The optimizer also accepts hints that give applications the flexibility to make tradeoffs between such factors as temporary space usage and performance.

See "Query Optimization" for more information about the query optimizer and indexing techniques.

Concurrency

TimesTen and IMDB Cache provide full support for shared databases. Options are available so users can choose the optimum balance between response time, throughput and transaction semantics for an application.

Read-committed isolation provides nonblocking operations and is the default isolation level. For databases with extremely strict transaction semantics, serializable isolation is available. These isolation levels conform to the ODBC standard and are implemented with optimal performance in mind. As defined by the ODBC standard, a default isolation level can be set for a TimesTen or IMDB Cache database, which can be dynamically modified for each connection at runtime.

For more information about managing concurrent operations in TimesTen and IMDB Cache, see "Concurrent Operations".

Automatic data aging

Data aging is an operation to remove data that is no longer needed. There are two general types of data aging: removing old data based on some time value or removing data that has been least recently used (LRU). For example, you can remove yesterday's price list, remove profiles and preferences of users who have logged out from the system, or remove detailed records that are more than 2 days old.

Two types of automatic data aging capability for TimesTen database tables and IMDB Cache data are available:

  • Time-based data aging based on timestamp values

  • Usage-based data aging based on the LRU algorithm

For more information, see "Implementing aging in your tables" in Oracle TimesTen In-Memory Database Operations Guide and "Implementing aging in a cache group" in Oracle In-Memory Database Cache User's Guide.

Globalization support

TimesTen and IMDB Cache provide globalization support for storing, retrieving, and processing data in native languages. Over 50 different national, multinational, and vendor-specific character sets including the most popular single-byte and multibyte encodings, plus Unicode, are supported as the database storage character set. The connection character set can be defined to enable an application running in a different encoding to communicate to the TimesTen or IMDB Cache database; character set conversion between the application and the database occurs automatically and transparently.

TimesTen and IMDB Cache offer linguistic sorting capabilities that handle the complex sorting requirements of different languages and cultures. More than 80 linguistic sorts are provided. They can be extended to enable the application to perform case-insensitive and accent-insensitive sorting and searches.

For more information, see "Globalization Support" in Oracle TimesTen In-Memory Database Operations Guide.

Transaction log monitoring

Like several other database systems, TimesTen and IMDB Cache have an API that allows applications to monitor update activities in order to generate actions outside the database. In TimesTen and IMDB Cache, this capability is provided by the Transaction Log API (XLA), which allows applications to monitor update records as they are committed and take various actions based on the detected updates. For example, an XLA application can apply the detected updates to another database, which could be TimesTen or a disk-based RDBMS. Another type of XLA application can notify subscribers that an update of interest has taken place.

TimesTen and IMDB Cache provide materialized views that can be used with XLA to enable notification of events described by SQL queries.

See "Event Notification" for more information about the Transaction Log API and materialized views.

Administration and utilities

TimesTen and IMDB Cache support typical database utilities such as interactive SQL, backup and restore, copy (copies data between different database systems), and migrate (a high speed copy for moving data between different versions of TimesTen or IMDB Cache).

Many administrative activities are available by using SQL extensions. TimesTen and IMDB Cache also use SQL extensions to set up replication, caching from an Oracle database and materialized views.

TimesTen built-in procedures and C language functions enable programmatic control over TimesTen operations and settings. TimesTen command-line utilities allow users to monitor the status of connections, locks, replication, and so on. Status can also be obtained using SQL SELECT queries on the system tables in the TimesTen schema. For example, the TimesTen SYS.SYSTEMSTATS table records many statistics that are useful for analyzing or debugging a TimesTen application.

For more information on TimesTen administration, see "TimesTen and IMDB Cache Administration".

Replication

TimesTen and IMDB Cache replication enable real-time data replication between servers for high availability and load sharing. Data replication configurations can be active-standby or active-active, using asynchronous or synchronous transmission, with conflict detection and resolution and automatic resynchronization after a failed server is restored.

See "Replication".

IMDB Cache

The Oracle In-Memory Database Cache creates a real-time, updatable cache for Oracle data. It offloads computing cycles from Oracle databases and enables responsive and scalable real-time applications. IMDB Cache loads a subset of Oracle tables into a cache database. It can be configured to propagate updates in both directions and to automate passthrough of SQL requests for uncached data. It automatically resynchronizes data after failures.

See "IMDB Cache".